[go: up one dir, main page]

WO2025208044A1 - Methods for cancer detection using molecular patterns - Google Patents

Methods for cancer detection using molecular patterns

Info

Publication number
WO2025208044A1
WO2025208044A1 PCT/US2025/022042 US2025022042W WO2025208044A1 WO 2025208044 A1 WO2025208044 A1 WO 2025208044A1 US 2025022042 W US2025022042 W US 2025022042W WO 2025208044 A1 WO2025208044 A1 WO 2025208044A1
Authority
WO
WIPO (PCT)
Prior art keywords
additional
individual
sequence representations
tumor
value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/US2025/022042
Other languages
French (fr)
Inventor
Anton VALOUEV
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guardant Health Inc
Original Assignee
Guardant Health Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guardant Health Inc filed Critical Guardant Health Inc
Publication of WO2025208044A1 publication Critical patent/WO2025208044A1/en
Pending legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B30/00—ICT specially adapted for sequence analysis involving nucleotides or amino acids
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
    • G16B40/20—Supervised data analysis
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B45/00—ICT specially adapted for bioinformatics-related data visualisation, e.g. displaying of maps or networks

Definitions

  • Cancer is a major cause of disease worldwide. Each year, tens of millions of people are diagnosed with cancer around the world, and more than half eventually die from it. In many countries, cancer ranks the second most common cause of death following cardiovascular diseases. Early detection is associated with improved outcomes for many cancers.
  • cancers are often detected by biopsies of tumors followed by analysis of cell markers or DNA extracted from cells. But more recently it has been proposed that cancers can also be detected from cell-free nucleic acids in body fluids, such as blood or urine. Such tests have the advantage that they are noninvasive and can be performed without identifying suspected cancer cells in biopsy. However, such tests are complicated by the fact that the amount of nucleic acids in body fluids is very low and that the nucleic acids that are present are heterogeneous in form (e.g., RNA and DNA, single-stranded and double-stranded, and various states of post-replication modification and association with proteins, such as histones).
  • body fluids such as blood or urine.
  • Figure 1 is a diagrammatic representation of an example computational architecture that implements one or more convolutional neural networks to identify samples obtained from subjects in which a tumor-related biological condition is present, according to one or more example implementations.
  • Figure 2 is a diagrammatic representation of an example computational architecture to generate image data derived from sample data and implementing a number of convolutional neural networks to analyze the image data for the detection of tumor-related biological conditions, according to one or more example implementations.
  • Figure 3 is a diagrammatic representation of an example computational architecture to implement one or more convolutional neural networks to detect a plurality of cancer types, according to one or more example implementations.
  • Figure 4 is a flow diagram of an example process to generate image data and implement one or more convolutional neural networks to analyze the image data for the detection of a tumor-derived biological condition, according to one or more implementations.
  • Figure 5 is a block diagram illustrating components of a machine, in the form of a computer system, that may read and execute instructions from one or more machine- readable media to perform any one or more methodologies described herein, in accordance with one or more example implementations.
  • the techniques described herein relate to a method including obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a plurality of groups of sequence representations that correspond to a plurality of genomic regions, individual groups of sequence representations of the plurality of groups of sequence representations being aligned with a discrete genomic region; computationally analyzing the individual groups of sequence representations to determine methylation rates of cytosine-guanine dinucleotides included in the individual groups of sequence representations; determining, based on the methylation rates, subsets of sequence representations from among the plurality of groups of sequence representations, individual subsets satisfying one or more methylation rate criteria; generating, based on individual subsets of sequence representations, one or more images for individual genomic regions of the plurality of genomic regions, individual images of the one or more images including a plurality of pixels, wherein individual pixels of the pluralit
  • the techniques described herein relate to a method including: obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected; generating first training images based on the first training sequence representations, individual first training images including a first plurality of pixels, wherein individual pixels of the first plurality of pixels comprise (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to methylation rate of at least a portion of the first training sequences representations, and (iii) a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value; obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected; generating second training images based on the second training sequence representations, individual second training images including a second plurality of pixels, wherein individual pixels of the second plurality of pixels comprise (i) a first additional training
  • the techniques described herein relate to a method wherein individual iterations of the plurality of iterations include: determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network; determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network; determining differences between the first weights and the second weights; and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights.
  • the techniques described herein relate to a method wherein the methylation rates of cytosine-guanine dinucleotides included in the individual sequence representations are determined by procedures that affect a first nucleobase differently from a second nucleobase.
  • the techniques described herein relate to a method wherein a logistic regression technique is implemented to determine the overall tumor indication based on the tumor indications determined by the plurality of convolutional neural networks.
  • the techniques described herein relate to a method wherein the probabilities of a tumor being present in the one or more subjects are analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
  • the techniques described herein relate to a method wherein the convolutional neural network includes a plurality of output layers with each output layer of the plurality of output layers corresponding to a cancer type of a plurality of cancer types; the convolutional neural network determines a plurality of probabilities of the plurality of cancer types being present in one or more subjects; and for individual output layers, the convolutional neural network generates an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in one or more subjects.
  • the techniques described herein relate to a method including computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
  • the techniques described herein relate to a method wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels.
  • the techniques described herein relate to a method including determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
  • the techniques described herein relate to a system wherein the memory stores additional computer-readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the individual subsets of sequence representations are computational analyzed to generate one or more additional images for the individual genomic regions, wherein the one or more additional images for the individual genomic regions includes first pixel values that comprise (i) first values that correspond to genomic locations within the individual genomic region and (ii) second values that correspond to an additional molecular characteristic of the individual sequence representations of the group of sequence representations.
  • the techniques described herein relate to a system wherein the one or more additional molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the subset of sequence representations corresponding to the individual genomic region, a length of the individual sequence representations of the subset of sequence representations corresponding to the individual genomic region, or a number of restriction enzyme cut sites in the individual sequence representations of the subset of sequence representations corresponding to the individual genomic region.
  • the techniques described herein relate to a system wherein the one or more methylation rate criteria correspond to at least a threshold number of methylated cytosine-guanine dinucleotides.
  • the techniques described herein relate to a system wherein the one or more methylation rate criteria correspond to no greater than a threshold number of methylated cytosine-guanine dinucleotides.
  • the techniques described herein relate to a system wherein the one or more samples are partitioned into a plurality of subsamples on the basis of methylate rate and the one or more methylation rate criteria correspond to a partition of a plurality of partitions into which the plurality of subsamples are divided.
  • the techniques described herein relate to a method including: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations, the group of sequence representations corresponding to one or more portions of a genomic region; computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations; generating, based on the group of sequence representations, one or more images that include a plurality of pixels, wherein individual pixels of the plurality of pixels include (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value; and providing the one
  • the techniques described herein relate to a method, including: computationally analyzing the sequencing data to determine a plurality of additional groups of additional sequence representations in relation to a plurality of additional genomic regions; computationally analyzing the plurality of additional groups of additional sequence representations to determine additional values of the one or more molecular characteristics for additional individual sequence representations of the plurality of additional groups of sequence representations; and generating a plurality of additional images based on the plurality of additional groups of sequence representations, wherein: the plurality of additional images include a plurality of additional pixels and individual additional pixels of the plurality of additional pixels include (i) an additional first value that corresponds to one or more additional genomic locations, (ii) an additional second value that corresponds to the one or more molecular characteristics, and (iii) an additional intensity value indicating an additional number of the additional sequence representations having the additional first value and the additional second value.
  • each additional image of the plurality of additional images is generated using information derived from additional sequence representations included in an additional group of sequence representations of the plurality of additional groups of sequence representations and the additional sequence representations are homologous with an additional genomic region.
  • the techniques described herein relate to a method, including: providing the plurality of additional images to a plurality of additional convolutional neural networks to determine a plurality of additional tumor indications related to a tumor being present in the one or more samples, wherein individual additional convolutional networks of the plurality of additional convolutional neural networks computationally analyze a portion of the plurality of additional images corresponding to a given genomic region to determine an additional tumor indication related to a tumor being present in the one or more subjects; and computationally analyzing the tumor indication and the plurality of additional tumor indications to determine an overall tumor indication related to a tumor being present in the one or more subjects.
  • the techniques described herein relate to a method, wherein the tumor indication and the plurality of additional tumor indications are computationally analyzed using a logistic regression technique to determine the overall tumor indication.
  • the techniques described herein relate to a method, wherein the probabilities of a tumor being present in the one or more subjects are computationally analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
  • the techniques described herein relate to a method, wherein: the convolutional neural network includes a plurality of output layers with each output layer of the plurality of output layers corresponding to a cancer type of a plurality of cancer types; the convolutional neural network determines a plurality of probabilities of the plurality of cancer types being present in one or more subjects; and for individual output layers, the convolutional neural network generates an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in one or more subjects.
  • the techniques described herein relate to a method, wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels.
  • the techniques described herein relate to a method, including: determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
  • the techniques described herein relate to a method, including: computationally analyzing the first image using a first convolutional neural network to determine a first tumor indication related to a tumor being present in one or more subjects; computationally analyzing the second image using a second convolutional neural network to determine a second tumor indication of a tumor being present in one or more subjects; and determining an overall tumor indication of a tumor being present in one or more subjects based on the first tumor indication and the second tumor indication.
  • the techniques described herein relate to a method, wherein the one or more molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the group of sequence representations, a length of the individual sequence representations of the group of sequence representations, or a number of restriction enzyme cut sites in the individual sequence representations of the group of sequence representations.
  • the techniques described herein relate to a method, including: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
  • the techniques described herein relate to a method, wherein the genomic region is included in a number of genomic regions that are enriched as part of a diagnostic test to determine the presence of tumors in subjects.
  • the techniques described herein relate to a method, including: obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected; generating first training images based on the first training sequence representations, individual first training images including a first plurality of pixels, wherein individual pixels of the first plurality of pixels include (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to the one or more molecular characteristics, and (iii) a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value; obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected; generating second training images based on the second training sequence representations, individual second training images including a second plurality of pixels, wherein individual pixels of the second plurality of pixels include (i) a first additional training value that corresponds to one or more genomic
  • the techniques described herein relate to a method, wherein individual iterations of the plurality of iterations include: determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network; determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network; determining differences between the first weights and the second weights; and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights.
  • the techniques described herein relate to a method including: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations, the group of sequence representations corresponding to one or more portions of a genomic region; computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations; generating, based on the group of sequence representations, one or more images that include a plurality of pixels, wherein individual pixels of the plurality of pixels include (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value; and providing the one
  • the techniques described herein relate to one or more computing apparatuses, including: one or more hardware processors; and memory storing computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising: computationally analyzing the sequencing data to determine a plurality of additional groups of additional sequence representations in relation to a plurality of additional genomic regions; computationally analyzing the plurality of additional groups of additional sequence representations to determine additional values of the one or more molecular characteristics for additional individual sequence representations of the plurality of additional groups of sequence representations; and generating a plurality of additional images based on the plurality of additional groups of sequence representations, wherein: the plurality of additional images include a plurality of additional pixels and individual additional pixels of the plurality of additional pixels include (i) an additional first value that corresponds to one or more additional genomic locations, (ii) an additional second value that corresponds to the one or more molecular characteristics, and (iii) an additional intensity value indicating an additional number
  • the techniques described herein relate to one or more computer apparatuses, wherein the memory stores additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
  • the techniques described herein relate to one or more computer apparatuses, wherein: the one or more images include a first image that corresponds to the genomic region and a second image that corresponds to the genomic region; the first image includes first pixel values that include (i) first values that correspond to genomic locations within the genomic region and (ii) second values that correspond to a first molecular characteristic of the individual sequence representations of the group of sequence representations; and the second image includes second pixel values that include (i) the first values that correspond to the genomic locations within the genomic region and (ii) additional second values that correspond to a second molecular characteristic of the individual sequence representations of the group of sequence representations.
  • the techniques described herein relate to one or more computer apparatuses, wherein the memory stores additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
  • the techniques described herein relate to one or more non-transitory computer-readable media storing computer-readable instructions that, when executed by one or more hardware processors, cause the one or more hardware processors to perform operations comprising : obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations, the group of sequence representations corresponding to one or more portions of a genomic region; computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations; generating, based on the group of sequence representations, one or more images that include a plurality of pixels, wherein individual pixels of the plurality of pixels include (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics
  • each additional image of the plurality of additional images is generated using information derived from additional sequence representations included in an additional group of sequence representations of the plurality of additional groups of sequence representations and the additional sequence representations are homologous with an additional genomic region.
  • the techniques described herein relate to one or more non-transitory computer-readable media, wherein the tumor indication and the plurality of additional tumor indications include probabilities of a tumor being present in the one or more subjects.
  • the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
  • the techniques described herein relate to one or more non-transitory computer-readable media, wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels.
  • the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
  • the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the first image using a first convolutional neural network to determine a first tumor indication related to a tumor being present in one or more subjects; computationally analyzing the second image using a second convolutional neural network to determine a second tumor indication of a tumor being present in one or more subjects; and determining an overall tumor indication of a tumor being present in one or more subjects based on the first tumor indication and the second tumor indication.
  • the techniques described herein relate to one or more non-transitory computer-readable media, wherein the one or more molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the group of sequence representations, a length of the individual sequence representations of the group of sequence representations, or a number of restriction enzyme cut sites in the individual sequence representations of the group of sequence representations.
  • the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
  • the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having no greater than a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
  • the techniques described herein relate to one or more non-transitory computer-readable media, wherein genomic locations that correspond to first values of the plurality of pixels correspond to an interval that includes a plurality of nucleotides. [0089] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein the genomic region is included in a number of genomic regions that are enriched as part of a diagnostic test to determine the presence of tumors in subjects.
  • the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected; generating first training images based on the first training sequence representations, individual first training images including a first plurality of pixels, wherein individual pixels of the first plurality of pixels include (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to the one or more molecular characteristics, and (iii) a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value; obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected; generating second training images based on the second training sequence representations, individual
  • “about” or “approximately” as applied to one or more values or elements of interest refers to a value or element that is similar to a stated reference value or element.
  • the term “about” or “approximately” refers to a range of values or elements that falls within 25%, 20%, 19%, 18%, 17%, 16%, 15%, 14%, 13%, 12%, 11 %, 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1 %, or less in either direction (greater than or less than) of the stated reference value or element unless otherwise stated or otherwise evident from the context (except where such number would exceed 100% of a possible value or element).
  • Adapter refers to a short nucleic acid (e.g., less than about 500 nucleotides, less than about 100 nucleotides, or less than about 50 nucleotides in length) that can be at least partially double-stranded and used to link to either or both ends of a given sample nucleic acid molecule.
  • Adapters can include sequences of nucleic acid primer binding sites to permit amplification of a nucleic acid molecule flanked by adapters at both ends, and/or a sequencing primer binding site, including primer binding sites for sequencing applications, such as various next-generation sequencing (NGS) applications.
  • NGS next-generation sequencing
  • the adapter is a Y-shaped adapter in which one end is blunt ended or tailed as described herein, for joining to a nucleic acid molecule, which is also blunt ended or tailed with one or more complementary nucleotides.
  • an adapter is a bell-shaped adapter that includes a blunt or tailed end for joining to a nucleic acid molecule to be analyzed.
  • Other examples of adapters include T-tailed and C-tailed adapters.
  • the threshold amount of homology can be at least about 90%, at least about 91 %, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, at least about 99.5%, or at least about 99.9%.
  • alignment can include determining whether a sequence representation has at least a threshold amount of homology with respect to a reference sequence.
  • Barcode As used herein, “barcode” or “molecular barcode” in the context of nucleic acids refers to a nucleic acid molecule comprising a sequence that can serve as a molecular identifier. For example, individual "barcode" sequences can be added to each DNA fragment during next -generation sequencing (NGS) library preparation so that each read can be identified and sorted before the final data analysis. In one or more examples, individual barcode sequences can be added to DNA fragments during NGS library preparation so that reads corresponding to each unique molecule included in a sample can be identified.
  • NGS next -generation sequencing
  • cancer type refers to a type or subtype of cancer defined, e.g., by histopathology. Cancer type can be defined by any conventional criterion, such as on the basis of occurrence in a given tissue (e.g., blood cancers, central nervous system (CNS), brain cancers, lung cancers (small cell and non-small cell), skin cancers, nose cancers, throat cancers, liver cancers, bone cancers, lymphomas, pancreatic cancers, bowel cancers, rectal cancers, thyroid cancers, bladder cancers, kidney cancers, mouth cancers, stomach cancers, breast cancers, prostate cancers, ovarian cancers, lung cancers, intestinal cancers, soft tissue cancers, neuroendocrine cancers, gastroesophageal cancers, head and neck cancers, gynecological cancers, colorectal cancers, urothelial cancers, solid state cancers, heterogeneous cancers, homogenous cancers), unknown primary
  • tissue e.g., blood cancers, central
  • Carrier Signal refers to any intangible medium that is capable of storing, encoding, or carrying transitory or non-transitory instructions for execution by a machine, and includes digital or analog communications signals or other intangible medium to facilitate communication of such instructions. Instructions may be transmitted or received over a network using a transitory or non- transitory transmission medium via a network interface device and using any one of a number of data transfer protocols.
  • Cell-free nucleic acid refers to nucleic acids not contained within or otherwise bound to a cell or, in some implementations, nucleic acids remaining in a sample following the removal of intact cells.
  • Cell-free nucleic acids can include, for example, all non-encapsulated nucleic acids sourced from a bodily fluid (e.g., blood, plasma, serum, urine, cerebrospinal fluid (CSF), etc.) from a subject.
  • a bodily fluid e.g., blood, plasma, serum, urine, cerebrospinal fluid (CSF), etc.
  • Cell-free nucleic acids include DNA (cfDNA), RNA (cfRNA), and hybrids thereof, including genomic DNA, mitochondrial DNA, circulating DNA, siRNA, miRNA, circulating RNA (cRNA), tRNA, rRNA, small nucleolar RNA (snoRNA), Piwi- interacting RNA (piRNA), long non-coding RNA (long ncRNA), and/or fragments of any of these.
  • Cell-free nucleic acids can be double-stranded, single-stranded, or a hybrid thereof.
  • a cell-free nucleic acid can be released into bodily fluid through secretion or cell death processes, e.g., cellular necrosis, apoptosis, or the like.
  • cell-free nucleic acids are released into bodily fluid from cancer cells, e.g., circulating tumor DNA (ctDNA). Others are released from healthy cells.
  • CtDNA can be non-encapsulated tumor-derived fragmented DNA.
  • a cell-free nucleic acid can have one or more epigenetic modifications, for example, a cell-free nucleic acid can be acetylated, 5-methylated, ubiquitylated, phosphorylated, sumoylated, ribosylated, and/or citrull inated.
  • cellular nucleic acids means nucleic acids that are disposed within one or more cells at least at the point a sample is taken or collected from a subject, even if those nucleic acids are subsequently removed as part of a given analytical process.
  • Classification Region refers to a genomic region that may show sequence-independent changes in neoplastic cells (e.g., tumor cells and cancer cells) or that may show sequence-independent changes in cfDNA from subjects having cancer relative to cfDNA from subjects in which cancer is not present.
  • sequence-independent changes include, but are not limited to, changes in methylation rate (increases or decreases), nucleosome distribution, CTCF binding, transcription start sites, and regulatory protein binding regions.
  • sequence-independent changes in a classification region can indicate the presence of a single form of cancer in a subject.
  • the classification region can include from about 10 nucleotides to about 10,000 nucleotides, from about 50 nucleotides to about 8000 nucleotides, from about 100 nucleotides to about 5000 nucleotides, from about 50 nucleotides to about 2000 nucleotides, from about 25 nucleotides to about 250 nucleotides, from about 50 nucleotides to about 200 nucleotides, or from about 75 nucleotides to about 150 nucleotides.
  • classification region can be a differentially methylated region.
  • DMR refers to a region of DNA, such as a region of a genome, having a detectably different degree of methylation or a different methylation state in at least one cell or tissue type relative to the degree of methylation in the same region of DNA from at least one other cell or tissue type; or having a detectably different degree of methylation in at least one cell or tissue type obtained from a subject having a disease or disorder relative to the degree of methylation in the same region of DNA in the same cell or tissue type obtained from a healthy subject.
  • Communications Network refers to one or more portions of a network that may be an ad hoc network, an intranet, an extranet, a virtual private network (VPN), a local area network (LAN), a wireless LAN (WLAN), a wide area network (WAN), a wireless WAN (WWAN), a metropolitan area network (MAN), the Internet, a portion of the Internet, a portion of the Public Switched Telephone Network (PSTN), a plain old telephone service (POTS) network, a cellular telephone network, a wireless network, a Wi-Fi® network, another type of network, or a combination of two or more such networks.
  • VPN virtual private network
  • LAN local area network
  • WLAN wireless LAN
  • WAN wide area network
  • WWAN wireless WAN
  • MAN metropolitan area network
  • PSTN Public Switched Telephone Network
  • POTS plain old telephone service
  • a network or a portion of a network may include a wireless or cellular network and the coupling may be a Code Division Multiple Access (CDMA) connection, a Global System for Mobile communications (GSM) connection, or other type of cellular or wireless coupling.
  • CDMA Code Division Multiple Access
  • GSM Global System for Mobile communications
  • CpG As used herein, “CpG” or “cytosine-guanine dinucleotide” refers to a cytosine-phosphate-guanine site within a nucleic acid molecule sequence such that a cytosine molecule is followed by a guanine molecule in a 5’ -> 3’ direction of the nucleic acid molecule sequence.
  • deoxyribonucleic Acid or Ribonucleic Acid refers to a natural or modified nucleotide which has a hydrogen group at the 2'-position of the sugar moiety.
  • DNA can include a chain of nucleotides comprising four types of nucleotide bases: adenine (A), thymine (T), cytosine (C), and guanine (G).
  • ribonucleic acid or “RNA” refers to a natural or modified nucleotide which has a hydroxyl group at the 2’-position of the sugar moiety.
  • nucleic acid sequencing data denotes any information or data that is indicative of the order and identity of the nucleotide bases (e.g., adenine, guanine, cytosine, and thymine or uracil) in a molecule (e.g., a whole genome, whole transcriptome, exome, oligonucleotide, polynucleotide, or fragment) of a nucleic acid such as DNA or RNA
  • sequence information obtained using all available varieties of techniques, platforms or technologies, including, but not limited to capillary electrophoresis, microarrays, ligation-based systems, polymerase-based systems, hybridization-based systems, direct or indirect nucleotide identification systems, pyrosequencing, ion- or pH-based detection systems, and electronic signature-based systems.
  • differentially methylated region refers to a region of DNA, such as a region of a genome, having a detectably different degree of methylation or different methylation state in at least one cell or tissue type relative to the degree of methylation in the same region of DNA from at least one other cell or tissue type; or having a detectably different degree of methylation in at least one cell or tissue type obtained from a subject having a disease or disorder relative to the degree of methylation in the same region of DNA in the same cell or tissue type obtained from a healthy subject.
  • a differentially methylated region can include a genomic region, such as a genomic, region corresponding to immune system function, that has a greater number of methylated nucleic acid molecules in a given sample due to a higher than expected turnover of cells related to the genomic region in an organ caused by the presence of a tumor in the organ.
  • a genomic region such as a genomic, region corresponding to immune system function
  • a differentially methylated region has a detectably lower degree of methylation (e.g., unmethylated region comprising unmethylated cytosines) in at least one cell or tissue type, such as at least one immune cell type, relative to the degree of methylation in the same region of DNA from at least one other cell or tissue type, such as other immune cell types and/or cell types that contribute to cfDNA in healthy individuals, or from the same cell or tissue type from a healthy subject.
  • driver mutation means a mutation that drives cancer progression.
  • epigenetic target regions refers to target regions that may show sequence-independent differences in different cell or tissue types (e.g., different types of immune cells) or in neoplastic cells (e.g., tumor cells and cancer cells) relative to normal cells; or that may show sequence- independent differences (i.e., in which there is no change to the nucleotide sequence, e.g., differences in methylation, nucleosome distribution, or other epigenetic features) in DNA, such as cfDNA, from different cell types or from subjects having cancer relative to DNA, such as cfDNA, from healthy subjects, or in cfDNA originating from different cell or tissue types that ordinarily do not substantially contribute to cfDNA (e.g., immune, lung, colon, etc.) relative to background cfDNA (e.g., cfDNA that originated from hematopoietic cells).
  • sequence-independent changes include, but are not limited to, changes in methylation (increases or decreases), nucleosome distribution, cfDNA fragmentation patterns, CCCTC-binding factor (“CTCF”) binding, transcription start sites (e.g., with respect to any one of more of binding of RNA polymerase components, binding of regulatory proteins, fragmentation characteristics, and nucleosomal distribution), and regulatory protein binding regions.
  • Epigenetic target region sets thus include, but are not limited to, hypermethylation variable target region sets, hypomethylation variable target region sets, and fragmentation variable target region sets, such as CTCF binding sites and transcription start sites.
  • loci susceptible to neoplasia-, tumor-, or cancer-associated focal amplifications and/or gene fusions may also be included in an epigenetic target region set because detection of a change in copy number by sequencing or a fused sequence that maps to more than one locus in a reference genome tends to be more similar to detection of exemplary epigenetic changes discussed above than detection of nucleotide substitutions, insertions, or deletions, e.g., in that the focal amplifications and/or gene fusions can be detected at a relatively shallow depth of sequencing because the ⁇ r detection does not depend on the accuracy of base calls at one or a few individual positions.
  • An epigenetic target region set is a set of two or more epigenetic target regions.
  • Immunotherapy refers to treatment with one or more agents that act to stimulate the immune system so as to kill or at least to inhibit growth of cancer cells, and preferably to reduce further growth of the cancer, reduce the size of the cancer and/or eliminate the cancer. Some such agents bind to a target present on cancer cells; some bind to a target present on immune cells and not on cancer cells; some bind to a target present on both cancer cells and immune cells. Such agents include, but are not limited to, checkpoint inhibitors, genetically engineered immune cells and/or antibodies, including natural antibodies and genetically engineered antibodies.
  • Checkpoint inhibitors are inhibitors of pathways of the immune system that maintain self-tolerance and modulate the duration and amplitude of physiological immune responses in peripheral tissues to minimize collateral tissue damage (see, e.g., Pardoll, Nature Reviews Cancer 12, 252-264 (2012)).
  • Example agents include antibodies against any of PD-1 , PD-2, PD-L1 , PD-L2, CTLA-40, 0X40, B7.1 , B7He, LAG3, CD137, KIR, CCR5, CD27, or CD40.
  • Other example agents include proinflammatory cytokines, such as IL-1 [3, IL-6, and TNF-a.
  • T-cells activated against a tumor such as T-cells designed to be activated by expressing a chimeric antigen receptor (CAR) targeting a tumor antigen and/or cell-surface protein engineered to be recognized by the T-cell.
  • CAR chimeric antigen receptor
  • Indel refers to a mutation that involves the insertion or deletion of nucleotides in the genome of a subject.
  • machine-readable medium refers to a component, device, or other tangible media able to store instructions and data temporarily or permanently and may include, but is not limited to, random-access memory (RAM), read-only memory (ROM), buffer memory, flash memory, optical media, magnetic media, cache memory, other types of storage (e.g., erasable programmable read-only memory (EEPROM)) and/or any suitable combination thereof.
  • RAM random-access memory
  • ROM read-only memory
  • buffer memory flash memory
  • optical media magnetic media
  • cache memory other types of storage
  • EEPROM erasable programmable read-only memory
  • machine- readable medium may be taken to include a single medium or multiple media (e.g., a centralized or distributed database, or associated caches and servers) able to store instructions.
  • machine-readable medium shall also be taken to include any medium, or combination of multiple media, that is capable of storing instructions (e.g., code) for execution by a machine and that when executed by one or more processors of the machine, cause the machine to perform any one or more of the methodologies described herein. Accordingly, a “machine-readable medium” refers to a single storage apparatus or device, as well as “cloud-based” storage systems or storage networks that include multiple storage apparatus or devices. The term “machine-readable medium” excludes signals per se.
  • methylation refers to addition of a methyl group to a nucleotide base in a nucleic acid molecule.
  • methylation refers to addition of a methyl group to a cytosine at a CpG site.
  • DNA methylation refers to addition of a methyl group to adenine, such as in N 6 -methyladenine.
  • DNA methylation is 5-methylation (modification of the 5th carbon of the 6-carbon ring of cytosine).
  • 5-methylation refers to addition of a methyl group to the 5C position of the cytosine to create 5-methylcytosine (5mC).
  • methylation comprises a derivative of 5mC.
  • Derivatives of 5mC include, but are not limited to, 5- hydroxymethylcytosine (5-hmC), 5-formylcytosine (5-fC), and 5-caryboxylcytosine (5- caC).
  • DNA methylation is 3C methylation (modification of the 3rd carbon of the 6-carbon ring of cytosine).
  • 3C methylation comprises addition of a methyl group to the 3C position of the cytosine to generate 3- methylcytosine (3mC).
  • Methylation rate refers to the probability, likelihood, or percentage that a given base (for example: cytosine residue in a CpG) is methylated on a DNA molecule at a particular genomic region analyzed in the sample.
  • the methylation rate may be applied to a defined region that comprises one or more potentially methylated bases.
  • the methylation rate refers to the percentage of CpG residues methylated in a DNA molecule.
  • the methylation rate refers to the percentage of CpG residues methylated in molecules aligned to particular genomic position or genomic region.
  • Methylation rate can be measured by a variety of methods including, but not limited to, either using bisulfite sequencing (any single base resolution like TAPS, EM-SEQ, etc.) or using partitioning (DNA molecule resolution), such as DNA methylation partitioning using methyl-binding antibodies or proteins. Methylation rate can be measured in different ways. One estimation can be by counting how many DNA fragments end up in each methylation dependent partition or by counting the number of converted CpGs per fragment in the case of bisulfite sequencing or any other base-level resolution sequencing methods, such as qPCR-based methods that can use either converted DNA or methyl- precipitated DNA.
  • the rate calculation can be normalized using a set of predefined genomic control regions with known methylation state (i.e., positive control regions and/or negative control regions) or spiked- in synthetic DNA with known methylation state, deriving rate-parametrized partition distributions and estimating the rate using a maximum likelihood approach.
  • methylation rate can be calculated by dividing or “normalizing” the count of methylated molecules corresponding to one or more genomic regions by the number of molecules present within the genomic control regions.
  • the methylation rate can be determined by measuring an abundance of sequencing reads that correspond to a portion of a genomic region. The portion of the genomic region can include a number of genomic locations of the genomic region for which at least a threshold number of sequencing reads overlap.
  • Methylation Status can refer to the presence or absence of methyl group on a DNA base (e.g., cytosine) at a particular genomic position in a nucleic acid molecule. It can also refer to the degree of methylation in a nucleic acid sequence (e.g., highly methylated, low methylated, intermediately methylated or unmethylated nucleic acid molecules). The methylation status can also refer to the number of nucleotides methylated in a particular nucleic acid molecule.
  • Mutation count refers to the number of somatic mutations in a whole genome or exome or targeted regions of a nucleic acid sample.
  • Neoplasm As used herein, the terms “neoplasm” and “tumor” are used interchangeably. They refer to abnormal growth of cells in a subject.
  • a neoplasm or tumor can be benign, potentially malignant, or malignant.
  • a malignant tumor is referred to as a cancer or a cancerous tumor.
  • next generation sequencing refers to sequencing technologies having increased throughput as compared to traditional Sanger- and capillary electrophoresis-based approaches, for example, with the ability to generate hundreds of thousands of relatively small sequencing reads at a time.
  • next generation sequencing techniques include, but are not limited to, sequencing by synthesis, sequencing by ligation, and sequencing by hybridization.
  • nucleic acid tags may be used to label different nucleic acid molecules or different nucleic acid samples or sub-samples.
  • Nucleic acid tags can be single-stranded, double-stranded, or at least partially double-stranded. Nucleic acid tags optionally have the same length or varied lengths. Nucleic acid tags can also include double-stranded molecules having one or more blunt-ends, include 5’ or 3’ single-stranded regions (e.g., an overhang), and/or include one or more other single-stranded regions at other locations within a given molecule. Nucleic acid tags can be attached to one end or to both ends of the other nucleic acids (e.g., sample nucleic acids to be amplified and/or sequenced).
  • Nucleic acid tags can be decoded to reveal information such as the sample of origin, form, or processing of a given nucleic acid.
  • nucleic acid tags can also be used to enable pooling and/or parallel processing of multiple samples comprising nucleic acids bearing different molecular barcodes and/or sample indexes in which the nucleic acids are subsequently being deconvolved by detecting (e.g., reading) the nucleic acid tags.
  • Nucleic acid tags can also be referred to as identifiers (e.g., molecular identifier, sample identifier).
  • nucleic acid tags can be used as molecular identifiers (e.g., to distinguish between different molecules or amplicons of different parent molecules in the same sample or sub-sample). This includes, for example, uniquely tagging different nucleic acid molecules in a given sample, or non-uniquely tagging such molecules.
  • tags i.e., molecular barcodes
  • endogenous sequence information for example, start and/or stop positions where they map to a selected reference sequence, a sub-sequence of one or both ends of a sequence, and/or length of a sequence
  • a polynucleotide is represented by a sequence of letters, such as “AGCTG,” it will be understood that the nucleotides are in 5’ -> 3’ order from left to right and that in the case of DNA, “A” denotes deoxyadenosine, “C” denotes deoxycytidine, “G” denotes deoxyguanosine, and “T” denotes deoxythymidine, unless otherwise noted.
  • the letters A, C, G, and T may be used to refer to the bases themselves, to nucleosides, or to nucleotides comprising the bases, as is standard in the art.
  • Probe refers to a polynucleotide comprising a functionality.
  • the functionality can be a detectable label (fluorescent), a binding moiety (biotin), or a solid support (a magnetically attractable particle or a chip).
  • Probes can include single-stranded DNA/RNA polynucleotides or double stranded DNA polynucleotides that hybridize to target nucleic acid sequences (e.g., SureSelect® probes, Agilent Technologies). Sequence capture using probes generally depends, in part, on the number of consecutive nucleotides in at least a portion of the target nucleic acid sequence that is complementary (or nearly complementary) to the sequence of the probe. In some examples, probes can correspond to driver mutations.
  • processing can be used interchangeably.
  • the terms refer to determining a difference, e.g., a difference in number or sequence.
  • a difference in number or sequence e.g., gene expression, copy number variation (CNV), indel, and/or single nucleotide variant (SNV) values or sequences can be processed.
  • CNV copy number variation
  • SNV single nucleotide variant
  • processor refers to any circuit or virtual circuit (a physical circuit emulated by logic executing on an actual processor) that manipulates data values according to control signals (e.g., "commands,” “op codes,” “machine code,” etc.) and which produces corresponding output signals that are applied to operate a machine.
  • a processor may, for example, be a CPU, a RISC processor, a CISC processor, a GPU, a DSP, an ASIC, a RFIC or any combination thereof.
  • a processor may further be a multi-core processor having two or more independent processors (sometimes referred to as "cores") that may execute instructions contemporaneously.
  • Quantitative measures refers to an absolute or relative measure.
  • a quantitative measure can be, without limitation, a number, a statistical measurement (e.g., frequency, mean, median, standard deviation, or quantile), or a degree or a relative quantity (e.g., high, medium, and low).
  • a quantitative measure can be a ratio of two quantitative measures.
  • a quantitative measure can be a linear combination of quantitative measures.
  • a quantitative measure may be a normalized measure.
  • Reference Sequence refers to a known sequence used for purposes of comparison with experimentally determined sequences.
  • a known sequence can be an entire genome, a chromosome, or any segment thereof.
  • a reference sequence can include at least about 20, at least about 50, at least about 100, at least about 200, at least about 250, at least about 300, at least about 350, at least about 400, at least about 450, at least about 500, at least about 1000, or more nucleotides.
  • a reference sequence can align with a single contiguous sequence of a genome or chromosome or can include non-contiguous segments that align with different regions of a genome or chromosome.
  • Example reference sequences include, for example, human genome reference sequences, such as, hG19 and hG38.
  • sample means anything capable of being analyzed by the methods and/or systems disclosed herein.
  • Sequencing refers to any of a number of technologies used to determine the sequence (e.g., the identity and order of monomer units) of a biomolecule, e.g., a nucleic acid such as DNA or RNA.
  • Single nucleotide variant or “SNV” means a mutation or variation in a single nucleotide that occurs at a specific position in the genome.
  • Somatic Mutation means a mutation in the genome that occurs after conception. Somatic mutations can occur in any cell of the body except germ cells and accordingly, are not passed on to progeny.
  • a subject can be an individual who has been diagnosed with having a cancer, is going to receive a cancer therapy, and/or has received at least one cancer therapy.
  • the subject can be in remission of a cancer.
  • the subject can be an individual who is diagnosed of having an autoimmune disease.
  • the subject can be a female individual who is pregnant or who is planning on getting pregnant, who may have been diagnosed of or suspected of having a disease, e.g., a cancer, an auto-immune disease.
  • Threshold refers to a predetermined value used to characterize experimentally determined values of the same parameter for different samples depending on their relation to the threshold.
  • tumor fraction refers to the estimate of the fraction of nucleic acid molecules derived from a tumor in a given sample.
  • the tumor fraction of a sample can be a measure derived from the max MAF of the sample or pattern of sequencing coverage of the sample or length of the cfDNA fragments in the sample or any other selected feature of the sample. In some instances, the tumor fraction of a sample is equal to the max MAF of the sample.
  • Cancer is usually caused by the accumulation of mutations within genes of an individual's cells, at least some of which result in improperly regulated cell division.
  • Such mutations can include single nucleotide variations (SNVs), gene fusions, insertions, transversions, translocations, and inversions. These mutations can also include copy number variations that correspond to an increase or a decrease in the number of copies of a gene within a tumor genome relative to an individual’s noncancerous cells.
  • An extent of mutations present in cell-free nucleic acids and an amount of mutated cell-free nucleic acids of a sample can be used as biomarkers to determine tumor progression, predict patient outcome, and refine treatment choices. In various examples, the extent of mutations present in cell-free nucleic acids can be indicated by tumor cells copy number and tumor fraction for a given sample.
  • cancer can be indicated by non-sequence modifications, such as methylation.
  • methylation changes in cancer include local gains of DNA methylation in the CpG islands at the TSS of genes involved in normal growth control, DNA repair, cell cycle regulation, and/or cell differentiation. This increased amount of methylation can be associated with an aberrant loss of transcriptional capacity of involved genes and occurs at least as frequently as point mutations and deletions as a cause of altered gene expression.
  • DNA methylation profiling can be used to detect aberrant methylation in DNA of a sample.
  • the DNA can correspond to certain genomic regions (“differentially methylated regions” or “DMRs”) that are normally hypermethylated or hypomethylated in a given sample type (e.g., cfDNA from the bloodstream) but which may show an abnormal degree of methylation that correlates to a neoplasm or cancer, e.g., because of unusually increased contributions of tissues to the type of sample (e.g., due to increased shedding of DNA in or around the neoplasm or cancer) and/or from extents of methylation of the genome that are altered during development or that are perturbed by disease, for example, cancer or any cancer-associated disease.
  • DMRs genomic regions
  • cfDNA from the bloodstream e.g., cfDNA from the bloodstream
  • image data is generated using sequencing data that includes sequence representations derived from nucleic acid molecules present in one or more samples obtained from subjects.
  • the image data can be generated from alphanumeric representations of sequences of nucleotides that correspond to nucleic acid molecules present in one or more samples.
  • the image data can also be generated using numerical values of molecular characteristics of the nucleic acid molecules present in one or more samples.
  • pixels included in the image data can each comprise a pair of values.
  • a first value of the pair of values of individual pixels can correspond to genomic positions of the nucleic acid molecules with respect to a reference sequence and a second value of the pair of values of individual pixels can correspond to values of one or more molecular characteristics of the nucleic acid molecules corresponding to the sequence representations.
  • the one or more molecular characteristics can include at least one of length of individual sequence representations, number of cytosine-guanine dinucleotides present in individual sequence representations, number of methylated cytosine-guanine dinucleotides present in individual sequence representations, or number of restriction enzyme cut sites associated with the individual sequence representations.
  • the convolutional neural network architecture can include multiple convolutional neural networks with individual convolutional neural networks being implemented to analyze images generated from sequence representations that are aligned with an individual genomic region.
  • the genomic regions being analyzed can correspond to genomic regions that are enriched in relation to one or more diagnostic assays.
  • the convolutional neural network architecture can include 20 convolutional neural networks.
  • the convolutional neural network architecture can include 200 convolutional neural networks.
  • the convolutional neural network architecture can include 2000 convolutional neural networks.
  • the output from the individual convolutional neural networks can then be aggregated and analyzed to determine an overall indication with respect to a tumor- related biological condition being present or absent in relation to one or more subjects.
  • implementations described herein can leverage the use of sequence representation data obtained from patient samples rather than relying on data generated by cameras or medical imaging equipment.
  • implementations described herein are able to analyze image data that is more naturally suited to processing by the internal arrangement of layers of convolutional neural network.
  • implementations described herein can provide more accurate results than in scenarios where sequencing data and/or molecular characteristic values are themselves analyzed by the convolutional neural networks.
  • the accuracy of indications of tumor-related biological conditions determined by the implementations is increased with respect to existing systems because information from many genomic regions can be analyzed using separate convolutional neural networks.
  • Non-limiting examples of such cancers include biliary tract cancer, bladder cancer, transitional cell carcinoma, urothelial carcinoma, brain cancer, gliomas, astrocytomas, breast carcinoma, metaplastic carcinoma, cervical cancer, cervical squamous cell carcinoma, rectal cancer, colorectal carcinoma, colon cancer, hereditary nonpolyposis colorectal cancer, colorectal adenocarcinomas, gastrointestinal stromal tumors (GISTs), endometrial carcinoma, endometrial stromal sarcomas, esophageal cancer, esophageal squamous cell carcinoma, esophageal adenocarcinoma, ocular melanoma, uveal melanoma, gallbladder carcinomas, gallbladder adenocarcinoma, renal cell carcinoma, clear cell renal cell carcinoma, transitional cell carcinoma, urothelial carcinomas, Wilms tumor, leukemia, acute lymphocytic leukemia (ALL
  • the example computational architecture 100 can include, at 102, determining sequence representations 104 corresponding to a reference sequence 106.
  • determining sequence representations 104 that correspond to the reference sequence 106 can include aligning the sequence representations 104 with the reference sequence 106.
  • individual sequence representations 104 can be aligned with the reference sequence 106 by determining an amount of homology between individual nucleotides of the sequence representations 104 and positions of the reference sequence 106. The amount of homology between a given sequence representation 104 and a portion of the reference sequence 106 can be determined using BLAST programs (basic local alignment search tools) and PowerBLAST programs (Altschul et al., J. Mol.
  • the sequence representations 104 can include alphanumeric representations of nucleic acid molecules derived from one or more samples.
  • the sequence representations 104 can include, for individual nucleic acids, data that corresponds to a string of letters that represents the respective chains of nucleotides that correspond to the individual nucleic acid molecules derived from one or more samples.
  • the sequence representations 104 can be stored in one or more data files.
  • the sequence representations 104 be stored in a FASTQ file that comprises a text-based sequencing data file format storing raw sequence data and quality scores.
  • the sequence representations 104 can be stored in a data file according to a binary base call (BCL) sequence file format.
  • BCL binary base call
  • the sequence representations 104 can be stored in a BAM file.
  • the sequence representations 104 can comprise at least about one gigabyte (GB), at least about 2 GB, at least about 3GB, at least about 4 GB, at least about 5 GB, at least about 8 GB, or at least about 10 GB.
  • sequence representations 104 can be referred to herein as a “read” or a “sequencing read.”
  • individual nucleic acid molecules derived from one or more samples can correspond to multiple sequence representations 104 as a result of the amplification of the individual nucleic acid molecules derived from one or more samples.
  • individual nucleic acid molecules derived from the one or more samples can correspond to a single sequence representation as a result of the absence of amplification of the individual nucleic acid molecules.
  • the one or more samples that are analyzed to derive the sequence representations 104 can include one or more biological fluid samples obtained from one or more subjects.
  • the one or more samples can include one or more blood-based sample obtained from one or more subjects.
  • the one or more samples can include one or more plasma samples obtained from the one or more subjects.
  • the one or more samples can include one or more tissue samples obtained from one or more subjects.
  • the one or more subjects providing the one or more samples analyzed to generate the sequence representations 104 can include one or more mammals.
  • the one or more subjects providing the one or more samples analyzed to generate the sequence representations 104 can include one or more humans.
  • the one or more subjects providing the one or more samples analyzed to generate the sequence representations 104 can include one or more non-human mammals.
  • the sequence representations 104 can include a number of individual sequence representations including a first sequence representation 110, a second sequence representation 112, a third sequence representation 114, and a fourth sequence representation 116.
  • the first sequence representation 110, the second sequence representation 112, and the third sequence representation 114 are aligned with a portion of the genomic region 108 such that start positions and stop positions of the individual sequence representations 110, 112, 114 include or are within the start position and the stop position of the genomic region 108.
  • a portion of the fourth sequence representation 116 is aligned with a portion of the genomic region 108 and an additional portion of the fourth sequence representation 116 is aligned with a portion of the reference sequence 106 that is outside of the genomic region 108.
  • the first sequence representation 110, the second sequence representation 112, and the third sequence representation 114 comprise aligned sequence representations 118 that comprise at least a subset of the sequence representations 104 having start positions and stop positions that include and/or are within the start position and the stop position of the genomic region 108.
  • the sequence representations 104 can be derived from sequencing data that is generated as part of one or more sequencing processes performed with respect to nucleic acids obtained from one or more samples.
  • the sequencing data can include and/or be used to generate position data 120 that corresponds to the aligned sequence representations 118.
  • the position data 120 can indicate start positions and stop positions for individual aligned sequence representations 118 in relation to positions of the reference sequence 106.
  • the position data 120 can indicate an offset of at least one of a start position or a stop position of the aligned sequence representations 118 in relation to at least one of a start position or a stop position of the genomic region 108.
  • the position data 120 can indicate a chromosome that is corresponding to the aligned sequence representations 118.
  • the computational architecture 100 can include, at 122, determining molecular characteristics 124 of the aligned sequence representations 118.
  • the molecular characteristics 122 can be determined by analyzing sequencing data corresponding to the aligned sequence representations 118.
  • the sequencing data of the aligned sequence representations 118 can be analyzed to determine lengths of the individual aligned sequence representations 118.
  • the length of an aligned sequence representation 118 can correspond to a number of nucleotides included in the individual aligned sequence representation 118.
  • the sequencing data of the aligned sequence representations 118 can be analyzed to determine a number of cytosine-guanine dinucleotides (CpGs) included in the aligned sequence representations 118.
  • CpGs cytosine-guanine dinucleotides
  • the molecular characteristics 124 can be determined by analyzing methylation data corresponding to the aligned sequence representations 118.
  • the sequencing data can include methylation data.
  • the methylation data can be determined by one or more nucleobase methylation state detection processes.
  • the methylation data can indicate modified nucleotides that include one or more methyl groups that are not present in unmodified forms of the nucleotides.
  • the methylation data can indicate modified cytosines.
  • the methylation data can indicate a number of restriction enzyme cut sites for individual aligned sequence representations 118.
  • the molecular characteristics 124 of the aligned sequence representations 118 can be determined by analyzing methylation data to determine a number of methylated cytosines in the individual aligned sequence representations 118 and/or a range of methylated cytosines in the individual aligned sequence representations 118.
  • the computational architecture 100 can include, at 126, generating image data 128 based on the position data 120 and the molecular characteristics 124.
  • the image data 128 can include a number of images, such as an example image 130.
  • the images 130 can include a plurality of pixels with each pixel having a first value indicated by a position along the x-axis and a second value indicated by a position along the y-axis.
  • the images 130 can have at least 500 pixels, at least 1000 pixels, at least 2000 pixels, at least 5000 pixels, at least 10,000 pixels, at least 25,000 pixels, at least 50,000 pixels, at least 75,000 pixels, at least 100,000 pixels, or more.
  • the first value of the pixels included in the images 130 can correspond to a genomic position.
  • the first values of the pixels in the images 130 along the x-axis can correspond to positions of the genomic region 108.
  • the first values of the pixels in the images 130 along the x-axis can correspond to individual positions of the genomic region 108.
  • the first values of the pixels of the images 130 along the x-axis can correspond to a group of positions of the genomic region 108. In these scenarios, the first values of the pixels of the images 130 along the x-axis can correspond to intervals of positions in the genomic region 108.
  • the interval corresponding to the first values of the pixels of the images 130 along the x-axis can correspond to every 2 positions of the genom ic region 108, every 3 positions of the genom ic region 108, every 4 positions of the genom ic region 108, every 5 positions of the genom ic region 108, every 6 positions of the genom ic region 108, every 7 positions of the genom ic region 108, every 8 positions of the genomic region 108, every 9 positions of the genomic region 108, every 10 positions of the genomic region 108, every 12 positions of the genomic region 108, every 14 positions of the genomic region 108, every 16 positions of the genomic region 108, every 18 positions of the genomic region 108, every 20 positions of the genomic region 108, every 25 positions of the genomic region 108, every 30 positions of the genomic region 108, every 40 positions of the genomic region 108, or every 50 positions of the genomic region 108.
  • the second values of the pixels of the images 130 along the y-axis can correspond to restriction enzyme cut sites for the aligned sequence representations 118. In various examples, the second values of the pixels of the images 130 along the y-axis can correspond to lengths of the aligned sequence representations 118.
  • the images 130 can include pixel values of the aligned sequence representations 118 such that individual pixels can be characterized by a pair of values.
  • the pair of values includes the x-axis value for an individual aligned sequence representation 118 that varies with the genomic positions of the individual aligned sequence representation 118 in conjunction with the genomic region 108 and a constant y-axis value that corresponds to the value of a specified molecular characteristic 124 of the individual aligned sequence representation 118.
  • individual lines included in the image 130 can have comprise pixels having different intensity values such that some intensity values of the pixels of the line have one value that corresponds to one number of the aligned sequence representations 118 overlapping with a set of genomic locations and a value of a molecular characteristic 124 and that other intensity values of the pixels of the line have another value that corresponds to another number of the aligned sequence representations 118 overlapping an additional set of genomic locations and having the same value of the molecular characteristic 124.
  • the convolutional neural networks included in the convolutional neural network architecture 140 can include a number of convolutional layers with individual layers being implemented to apply one or more convolutional filters to portions of an image. The values of the pixels can be modified according to the values of the filter to generate output values for a convolutional layer that can correspond to an output image.
  • the convolutional neural networks included in the convolutional neural network architecture 140 can identify one or more features within the images included in the image data 128. For example, individual convolutional layers of the convolutional neural network architecture 140 can generate feature maps based on the data input to the individual convolutional layer.
  • the convolutional neural network architecture 140 can include one or more rectified linear unit (ReLLI) layers that apply one or more rectification functions to the feature maps generated by the convolutional layers.
  • ReLLI rectified linear unit
  • the convolutional neural networks of the convolutional neural network architecture 140 can include one or more pooling layers.
  • the convolutional neural networks included in the convolutional neural network architecture 140 can include one or more Max Pooling layers.
  • the pooling layers of the convolutional neural network architecture 140 can reduce a size of dimensions of the feature maps produced by the convolutional layers. That is, as data corresponding to an image moves through layers of the convolutional neural network architecture 140, the number of pixels used to represent the image is reduced.
  • a flattening process can be performed prior to passing the modified feature maps to the output layers. The flattening process can generate onedimensional vectors from two-dimensional image data.
  • the convolutional neural networks of the convolutional neural network architecture 140 can include one or more output layers that correspond to one or more classifications.
  • the one or more classifications can be indicated by one or more probabilities determined by the convolutional neural networks of the convolutional neural network architecture 140 for the one or more classifications.
  • the output layers of the convolutional neural networks of the convolutional neural network architecture 140 can include fully-connected layers.
  • the output layers of the convolutional neural networks of the convolutional neural network architecture 140 can include one or more Softmax layers that include a number of nodes that apply a Softmax activation function to generate output probabilities.
  • the convolutional neural network architecture 140 can analyze the image data 128 that corresponds to a number of genomic regions to determine a tumor indication 142.
  • the tumor indication 142 can include at least one of a tumor fraction or a tumor burden.
  • the tumor indication 142 can indicate a binary determination that a tumor is present in a subject or that a tumor is not present in a subject.
  • the tumor indication 142 can indicate probabilities of a tumor being present in a subject.
  • the tumor indication 142 can correspond to an amount of progression of a tumor-related biological condition or an amount of regression of a tumor-related biological condition.
  • the tumor indication 142 can correspond to a responsiveness of a subject to one or more treatments. In one or more further examples, the tumor indication 142 can indicate a stage of cancer present in one or more subjects. In still other examples, the tumor indication 142 can correspond to one or more types of cancer present in the one or more subjects.
  • the convolutional neural network architecture 140 can implement a plurality of convolutional neural networks that individual generate a tumor indication that is combined into an overall tumor indication that is output by the convolutional neural network architecture 140 as the tumor indication 142. In one or more illustrative examples, the individual tumor indications produced by the plurality of convolutional neural networks can be combined using at least one of max pooling techniques, average pooling techniques, or one or more linear functions.
  • the one or more nucleobase methylation state detection processes that are used to generate one or more of the molecular characteristics 124 can include one or more chemical processes and/or biochemical processes that impact a first type of nucleotide differently than a second type of nucleotide.
  • the one or more nucleobase methylation state detection processes implemented to generate one or more of the molecular characteristics 124 can include one or more reactions that cause at least one atomic and/or molecular moiety of the first type of nucleotide to be modified in a manner that is different from the manner in which the one or more reactions affect the second type of nucleotide.
  • the impact of the one or more nucleobase methylation state detection processes on a given type of nucleotide can be based on one or more previous modifications to the given type of nucleotide in relation to an unmodified form of the given type of nucleotide. That is, in various examples, a molecule corresponding to a given type of nucleotide may have been modified before being subjected to the one or more nucleobase methylation state detection processes. To illustrate, before being subjected to the one or more nucleobase methylation state detection processes, nucleotides of nucleic acid molecules derived from one or more samples can be modified due to mutations caused by the presence of a tumor in a subject.
  • the one or more nucleobase methylation state detection processes can modify the first type of nucleotide or the second type of nucleotide such that the nucleobase pairing of the first type of nucleotide or the second type of nucleotide is altered.
  • the one or more nucleobase methylation state detection processes implemented to generate the molecular characteristics 124 can be performed on nucleic acid molecules included in one or more samples used to generate the number of sequence representations 104.
  • the one or more nucleobase methylation state detection processes can modify a first type of nucleotide of the nucleic acid molecules in a first manner and one or more additional types of nucleotides of the nucleic acid molecules in a second manner.
  • the one or more nucleobase methylation state detection processes can modify at least one of cytosines, guanines, thiamines, or adenines differently than at least one other of cytosines, guanines, thiamines, or adenines.
  • the one or more nucleobase methylation state detection processes can modify cytosines differently than guanines, thiamines, or adenines.
  • the one or more nucleobase methylation state detection processes can modify cytosines such that the modified cytosines no longer pair with guanines.
  • the one or more nucleobase methylation state detection processes may not modify 5-methylcytosines and/or 5- hydroxymethylcytosines of nucleic acid molecules derived from one or more samples In this way, the one or more nucleobase state detection processes can be used to differentiate cytosines that have been previously modified to include a 5-methyl group versus previously unmodified cytosines.
  • the one or more nucleobase methylation state detection processes can include at least one of sodium bisulfite conversion and sequencing, Tet-assisted bisulfite sequencing (TAB-Seq), differential enzymatic cleavage, one or more single molecule sequencing methods, such as nanopore DNA sequencing, oxidative bisulfite (Ox-BS) conversion, APOBEC-coupled epigenetic (ACE) conversion, or direct methylation sequencing (DM-Seq).
  • TAB-Seq Tet-assisted bisulfite sequencing
  • ACE APOBEC-coupled epigenetic
  • DM-Seq direct methylation sequencing
  • the one or more nucleobase methylation state detection processes can include one or more processes that separate nucleic acid molecules based on amounts of nucleotides of the nucleic acid molecules that have been previously modified.
  • the one or more nucleobase methylation state detection processes can determine a methylation rate for one or more regions of the nucleic acid molecules derived from one or more samples.
  • the one or more nucleobase methylation state detection processes can separate nucleic acid molecules included in one or more samples based on amounts of methylated cytosines included in CG regions of individual nucleic acid molecules.
  • the one or more nucleobase methylation state detection processes can separate the nucleic acid molecules derived from one or more samples into a plurality of groups of nucleic acid molecules with individual groups of nucleic acid molecules corresponding to respective amounts of methylated cytosines of the nucleic acid molecules.
  • the one or more nucleobase methylation state detection processes can include at least one of partitioning of nucleic acid molecules included in the one or more samples based on a strength of binding of the individual nucleic acid molecules to methyl binding domain (MBD) and, optionally, treatment with methylation sensitive restriction enzyme (MSRE) and/or methylation dependent restriction enzyme (MDRE).
  • a strength of binding of nucleic acid molecules to MBD can be determined by subjecting the nucleic acids to a series of washes having different concentrations of MBD.
  • the nucleobase methylation state detection processes can include one or more sequencing processes.
  • the nucleobase methylation state detection processes can include whole genome bisulfite sequencing, reduced representation bisulfite sequencing, targeted bisulfite sequencing, extended- representation bisulfite sequencing, or one or more combinations thereof.
  • whole genomic bisulfite sequencing can be performed according to the techniques described in T.
  • targeted bisulfite sequencing can be performed according to techniques described in D.A. Moser et al., “Targeted bisulfite sequencing: A novel tool for the assessment of DNA methylation with high sensitivity and increased coverage,” Psychoneuroendocrinology, 120:1 -8, 2020 and/or E. Leitao et al., “Locus-specific DNA methylation analysis by targeted deep bisulfite sequencing,” Methods Mol Biol, 1767:351-66, 2018.
  • extended-representation bisulfite sequencing can be performed according to techniques described in Shareef, S.J., Bevill, S.M., Raman, A.T. et al. Extended- representation bisulfite sequencing of gene regulatory elements in multiplexed samples and single cells. Nat Biotechnol 39, 1086-1094 (2021 ).
  • first sequence representations 210 can be produced with respect to the first genomic region 204
  • second sequence representations 212 can be produced with respect to the second genomic region 206
  • third sequence representations 214 can be produced with respect to the third genomic region 208.
  • the sequence representations 210, 212, 214 can correspond to nucleic acid molecules present in one or more samples obtained from one or more subjects.
  • the sequence representations 210, 212, 214 can correspond to nucleic acid molecules present in one or more samples provided by a subject that is being tested for a tumor-related biological condition.
  • the sequence representations 210, 212, 214 can include or correspond to sequencing reads produced in conjunction with one or more sequencing processes performed with respect to the one or more samples.
  • the first sequence representations 210 can have first molecule characteristics 216 and first position data 218, the second sequence representations 212 can have second molecular characteristics 220 and second position data 222, and the third sequence representations 214 can have third molecular characteristics 224 and third position data 226.
  • the molecular characteristics 216, 220, 224 can include at least one of length of the respective sequence representations 210, 212, 214; number of CpGs present in the respective sequence representations 210, 212, 214; number of methylated CpGs present in the respective sequence representations 210, 212, 214; or number of restriction enzyme cut sites for the respective sequence representations 210, 212, 214.
  • the position data 218, 222, 226 can indicate locations of the respective sequence representations 210, 212, 214 in relation to a reference genome. In one or more examples, the position data 218, 222, 226 can indicate at least one of start positions or stop positions of the respective sequence representations 210, 212, 214. Additionally, the position data 218, 222, 226 can indicate a chromosome corresponding to the respective sequence representations 210, 212, 214.
  • the position data 218, 222, 226 can indicate, for the respective sequence representations 210, 212, 214, an offset with respect to at least one of start positions of the genomic regions 204, 206, 208; stop positions of the genomic regions 204, 206, 208; start position of a chromosome; or a stop position of a chromosome.
  • the individual pixels can also have intensity values that correspond to a number of the sequence representations 210, 212, 214 that overlap with the given position data and molecular characteristic values of the individual pixels.
  • the intensity values for individual pixels included in images of the first image data 228, the second image data 230, and the third image data 232 can correspond to normalized intensity values.
  • the intensity values for pixel values of images included in the image data 228, 230, 232 can be normalized in relation to a maximum pixel intensity value for individual images included in the image data 228, 230, 232.
  • the intensity values for the pixels of images included in the image data 228, 230, 232 can include a logarithmic transformation of a normalized pixel value.
  • individual normalized pixel values can correspond to a number of sequence representations having the position value and molecular characteristic value for the pixel in relation to the number of sequence representations that correspond to a given genomic region.
  • the normalized pixel values can be determined in relation to quantitative measures, such as counts, of the number of sequence representations corresponding to a number of control regions.
  • a pixel of an image included in the first image data 228 can have a first value corresponding to a position of a nucleotide included in one or more of the first sequence representations 210 and a second value corresponding to a molecular characteristic of the one or more first sequence representations 210.
  • the normalized intensity value for the pixel can be determined by a logarithmic transformation of a ratio of the number of first sequence representations 210 having the first value and the second value and the total number of the first sequence representations 210.
  • the first image data 228 can include multiple images with individual images of the multiple images being generated based on different molecular characteristics.
  • the first image data 228 can include a first image that is generated using values of a first molecular characteristic and a second image that is generated using values of a second molecular characteristic.
  • the x- values of the pixels in the first image and the second image can be the same and the y- values for the pixels in the first image and the second image can be different.
  • the first image data 228 can include a first image that includes pixels with x-values based on the first position data 218 and y-values that correspond to lengths of the first sequence representations 210 and a second image that includes pixels with x-values based on the first position data 218 and y-values that correspond to numbers of CpGs of the first sequence representations 210.
  • the second image data 230 and the third image data 232 can include multiple images that are generated based on values of different molecular characteristics.
  • one or more first tumor indications produced by the one or more first convolutional neural networks corresponding to a first molecular characteristic can be combined using one or more pooling techniques with one or more second tumor indications produced by one or more second convolutional neural networks corresponding to a second molecular characteristic.
  • the pooling layer can be trained using binary loss.
  • one or more first convolutional neural networks corresponding to a first molecular characteristic can generate multiple first outputs with each first output corresponding to an individual cancer type of a plurality of cancer types and one or more second convolutional neural networks corresponding to a second molecular characteristic can generate multiple second outputs with each second output corresponding to an individual cancer type of the plurality of cancer types.
  • the first outputs and the second outputs can be combined using one or more pooling techniques with training of the computational architecture 200 being performed using one or more softmax techniques.
  • the first image data 228, the second image data 230, and the third image data 232 can be provided to a computational system 234.
  • the computational system 234 can include multiple convolutional neural networks.
  • the individual convolutional neural networks included in the computational system 234 can analyze one of the first image data 228, the second image data 230, or the third image data 232.
  • the computational system 234 can include a first region convolutional neural network 236, a second region convolutional neural network 238, and a third region convolutional neural network 240.
  • the first region convolutional neural network 236 can analyze the first image data 228 to determine a first tumor indication 242 that corresponds to the first genomic region 204.
  • the second region convolutional neural network 238 can analyze the second image data 230 to determine a second tumor indication 244 that corresponds to the second genomic region 206.
  • the third region convolutional neural network 240 can analyze the third image data 232 to determine a third tumor indication 246.
  • the tumor indications 242, 244, 246 can individually indicate at least one of a tumor fraction, a tumor burden, a probability of a tumor-related biological condition being present in a subject, a progression of a tumor-related biological condition, a regression of a tumor- related biological condition, presence or absence of a tumor-related biological condition, or responsiveness to treatment for a tumor-related biological condition.
  • the output of the individual region convolutional neural networks 236, 238, 240 can include a logit value that corresponds to the tumor indications 242, 244, 246.
  • the computational system 234 can determine an overall tumor indication based on individual tumor indications generated by the region convolutional neural networks included in the computational system 234.
  • the overall tumor indication can be a combination of the first tumor indication 242, the second tumor indication 244, and the third tumor indication 246.
  • the tumor indications 242, 244, 246 can be analyzed according to at least one of one or more machine learning techniques or one or more statistical techniques to determine the overall tumor indication.
  • the overall tumor indication can be determined by using a logistic regression technique in relation to the first tumor indication 242, the second tumor indication 244, and the third tumor indication 246.
  • one or more non-linear computational methods can be applied to determine the overall tumor indication in relation to the first tumor indication 242, the second tumor indication 244, and the third tumor indication 246.
  • at least one of one or more random forests methods or one or more boosting tree methods can be implemented to determine the overall tumor indication in relation to the first tumor indication 242, the second tumor indication 244, and the third tumor indication 246.
  • labeled training data can include a first indication for position data and molecular characteristic data corresponding to sequence representations produced in relation to samples obtained from first training subjects in which a tumor- related biological condition is present and a second indication for position data and molecular characteristic data corresponding to sequence representations produced in relation to samples obtained from second training subjects in which a tumor-related biological condition is not present.
  • the first region convolutional neural network 236 can undergo a first training process based on the first training data corresponding to the first genomic region 204. Additionally, the second region convolutional neural network 238 can undergo a second training process based on the second training data corresponding to the second genomic region 206. Further, the third region convolutional neural network 240 can undergo a third training process based on the third training data.
  • the first training data, the second training data, and the third training data can include position data and molecular characteristic data for training sequence representations that correspond to the genomic regions 204, 206, 208 and that are generated in relation to the training samples. The training data can also include images generated from the position data and molecular characteristic data corresponding to the genomic regions 204, 206, 208 and obtained from the first training subjects and the second training subjects.
  • the training of the region convolutional neural networks 236, 238, 240 can include performing one or more feature extraction operations and one or more classification operations.
  • Feature extraction operations can include identifying one or more variables and/or one or more sets of variables that can be used to make one or more predictions based on a set of input data.
  • the feature extraction can determine relationships between one or more variables included in the training data and determine one or more measures of correlations between variables and/or groups of variables included in the training data.
  • the classification operations can include classifying one or more pieces of information included in the training data according to one or more categories.
  • a training process to generate the computational models of the region convolutional neural networks 236, 238, 240 can include at least 5 iterations, at least 10 iterations, at least 25 iterations, at least 50 iterations, at least 100 iterations, at least 500 iterations, at least 1000 iterations, or more
  • the one or more computational models of the region convolutional neural networks 236, 238, 240 can be produced after the feature extraction, classification, and prediction operations produce computational models that satisfy one or more performance criteria, such as one or more convergence criteria or one or more accuracy criteria.
  • the training process for the region convolutional neural networks 236, 238, 240 can include a forward phase and a backward phase.
  • inputs pass through the layers of the region convolutional neural networks 236, 238, 240 to produce an output.
  • backward phase gradients based on the outputs from the forward phase are backpropagated and weights of the layers are modified.
  • Data generated by one or more layers of the region convolutional neural networks 236, 238, 240 during the forward phase can be cached for later use in the backward phase of the training process.
  • the weights and/or biases of the individual layers of the region convolutional neural networks 236, 238, 240 can be determined using stochastic gradient descent techniques.
  • FIG. 3 is a diagrammatic representation of an example computational architecture 300 to implement one or more convolutional neural networks to detect a plurality of cancer types, according to one or more example implementations.
  • the computational architecture 300 can include a computing system 302 that analyzes image data generated from sequencing data and molecular characteristics data to determine an indication of a cancer type present in one or more subjects.
  • the computing system 302 can perform at least a portion of the operations described with respect to Figure 1 and Figure 2.
  • the computing system 302 can include one or more computing devices 304.
  • the one or more computing devices 304 can include at least one of one or more desktop computing devices, one or more mobile computing devices, or one or more server computing device.
  • At least a portion of the one or more computing devices 304 can be included in a remote computing environment, such as a cloud computing environment.
  • the operations executed by the computing system 302 can be performed by, controlled by, and/or maintained by a single organization. In one or more additional examples, the operations executed by the computing system 302 can be performed by, controlled by, and/or maintained by multiple organizations.
  • the region convolutional neural networks 306 can perform first output computations 308 that produces a first cancer type output 310.
  • the region convolutional neural networks 306 can also perform second output computations 312 that produces a second cancer type output 314.
  • the region convolutional neural networks 306 can perform third output computations 316 that generates a third cancer type output 318.
  • each output 310, 314, 318 can indicate a probability of a given cancer type being present in subjects.
  • the first output computations 308, the second output computations 312, and the third output computations 316 can be performed using one or more output layers of the region convolutional neural networks 306.
  • the first cancer type output 310, the second cancer type output 314, and the third cancer type output 318 can be produced using one or more softmax layers with each output 310, 314, 318 being determined using individual sets of weights corresponding to individual outputs 310, 314, 318.
  • the third output computations 316 can be performed using at least one of sequencing data, methylation data, genomic position data, or molecular characteristics data that is produced by analyzing at least one of sequence representations or molecular data of nucleic acid molecules that align with the genomic region and that is derived from samples obtained from subjects in which the third cancer type is present.
  • the training of the region convolutional neural network 306 with respect to different cancer types can be performed using batches of training data corresponding to the individual cancer types.
  • the training process to perform the output computations 308, 312, 316 of the region convolutional neural networks 306 can correspond to the training process for the region convolutional neural networks 236, 238, 240 included in the computational system 234 described in relation to Figure 2.
  • the computing system 302 can, at 326, analyze the cancer type outputs to determine a final cancer type output 328 for the genomic region associated with the region convolutional neural network 306.
  • the computational system 302 can analyze the first cancer type output 310, the second cancer type output 314, and the third cancer type output 318 to determine the final cancer type output 328.
  • the computing system 302 can implement one or more argmax functions based on the first cancer type output 310, the second cancer type output 314, and the third cancer type output 318 to determine the final cancer type output 328.
  • the final cancer type output 328 produced by the computing system 302 can include at least one of a tumor fraction, a tumor burden, an indication of tissue of origin, a probability of at least one type of cancer being present in one or more subjects, an indication of the presence of at least one type of cancer, an indication of the absence of at least one type of cancer, an indication of progression of at least one type of cancer, an indication of regression of at least one type of cancer, or an indication of responsiveness of one or more subjects to a treatment provided in relation to at least one type of cancer.
  • At least a portion of the individual region convolutional neural networks of the computing system 302 can include a different number of output layers corresponding to different cancer types for the individual genomic regions associated with the respective region convolutional neural networks.
  • the computing system 302 can aggregate the cancer type outputs from the individual region convolutional neural networks to determine an overall cancer type output for one or more subjects.
  • at least one of one or more machine learning models or one or more statistical techniques can be implemented to determine an overall cancer type output for one or more subjects based on final cancer type outputs 328 from a plurality of region convolutional neural networks corresponding to different genomic regions.
  • a logistic regression model can be implemented by the computing system 302 to aggregate final cancer type outputs 328 from a plurality of region convolutional neural networks to determine an overall cancer type output for one or more subjects.
  • Figure 4 is a flow diagram of an example process 400 to generate image data and implement one or more convolutional neural networks to analyze the image data for the detection of a tumor-derived biological condition, according to one or more implementations.
  • the process 400 can include obtaining sequencing data indicating a number of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects.
  • the sequencing data can include sequencing reads generated by one or more sequencing operations performed with respect to the nucleic acid molecules.
  • the sequencing data can also include methylation data indication one or more methylated CpGs present in the nucleic acid molecules.
  • the process 400 can include, at 404, determining, based on the sequencing data, a group of sequence representations that are aligned with respect to one or more portions of a genomic region.
  • the genomic region can be included in a number of genomic regions that are enriched as part of a diagnostic test to determine the presence of tumors in subjects.
  • the process 400 can also include, at 406, determining, based on the group of sequence representations, values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations.
  • the one or more molecular characteristics can include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the group of sequence representations.
  • the one or more molecular characteristics can also include a number of methylated cytosine-guanine dinucleotides present in an individual sequence representation of a group of sequence representations.
  • the one or more molecular characteristics can include a length of the individual sequence representations of the group of sequence representations.
  • the one or more molecular characteristics can include a number of restriction enzyme cut sites in the individual sequence representations of the group of sequence representations. in still other examples, the one or more molecular characteristics can include a number of sequence representations derived from a sample after one or more sequencing operations are performed that correspond to a given nucleic acid included in the sample. [00207] Further, at 408, the process 400 can include generating, based on the group of sequence representations, one or more images that include a plurality of pixels. Individual pixels of the plurality of pixels can comprise a first value that corresponds to a genomic location within the genomic region and a second value that corresponds to the one or more molecular characteristics.
  • the genomic locations that correspond to the first values of the plurality of pixels can correspond to an interval that comprises a plurality of nucleotides.
  • individual pixels of the plurality of pixels can include an intensity value indicating a number of the sequence representations included in the group of sequence representations having the first value and the second value. Intensity values of the plurality of pixels can increase as the number of sequence representation having the first value and the second value increases.
  • the intensity values of the plurality of pixels can be normalized based on a maximum intensity value for the plurality of pixels. In one or more illustrative examples, intensity value for the individual pixels of the plurality of pixels can be determined by determining a logarithmic transformation of a normalized pixel value. The normalized pixel value can correspond to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are aligned with respect to one or more control genomic regions.
  • the group of sequence representations used to generate the one or more images can be determined by analyzing the one or more molecular characteristics with respect to one or more criteria. For example, determining the group of sequence representations can include analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region.
  • the process 400 can include providing the one or more images to a convolutional neural network.
  • the convolutional neural network can computationally analyze the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
  • the convolutional neural network can determine multiple tumor indications that each correspond to an individual cancer type of a plurality of cancer types, in these scenarios, the multiple tumor indications can be determined by one or more output layers of the convolutional neural network.
  • the convolutional neural network can determine a plurality of probabilities of the plurality of cancer types being present in the one more subjects.
  • the plurality of additional images can be provided to a plurality of additional convolutional neural networks.
  • the plurality of additional convolutional neural networks can determine a plurality of additional tumor indications related to a tumor being present in the one or more samples.
  • Individual additional convolutional neural networks of the plurality of additional convolutional neural networks can analyze a portion of the plurality of additional images corresponding to a given genomic region to determine an additional tumor indication related to a tumor being present in the one or more subjects.
  • the group of tumor indications generated by the individual convolutional neural networks can be analyzed to determine an overall tumor indication related to a tumor being present in the one or more subjects.
  • multiple images can be generated for a genomic region with each image corresponding to a different molecular characteristic.
  • the one or more images can include a first image that corresponds to the genomic region and a second image that corresponds to the genomic region.
  • the first image can include first pixel values that comprise (i) first values that correspond to genomic locations within the genomic region and (ii) second values that correspond to a first molecular characteristic of the individual sequence representations of the group of sequence representations.
  • the second image can include second pixel values that comprise (i) the first values that correspond to the genomic locations within the genomic region and (ii) additional second values that correspond to a second molecular characteristic of the individual sequence representations of the group of sequence representations.
  • a plurality of iterations of a training process can be performed for the convolutional neural network to determine weights of layers of the convolutional neural network.
  • Individual iterations of the plurality of iterations of the training process for the convolutional neural network can include determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network.
  • individual iterations of the plurality of iterations of the training process for the convolutional neural network can include determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network.
  • the training process for the convolutional neural network can include determining differences between the first weights and the second weights and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights.
  • a training process for the convolutional neural network can be performed using samples from subjects in which cancer is present and samples from subjects in which cancer is not detected, during one or more iterations of the training process, a loss is calculated for individual sets of training data that include a portion of the training data derived from samples obtained from subjects in which cancer is present and a portion of the training data derived from samples obtained from subjects in which cancer is not detected. Weights for the convolutional neural network are updated based on gradients that are determined from differences in the loss values between iterations of the training process.
  • Individual pixels of the plurality of pixels can include (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value.
  • the method can include providing the one or more images to a convolutional neural network.
  • the convolutional neural network can computationally analyze the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
  • the techniques described herein relate to one or more computing apparatus that include: one or more hardware processors; and memory storing computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects.
  • the memory can also store additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations.
  • the group of sequence representations can correspond to one or more portions of a genomic region.
  • the memory can store additional computer- readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations and generating, based on the group of sequence representations, one or more images that include a plurality of pixels.
  • Individual pixels of the plurality of pixels can comprise (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value.
  • the memory can store additional computer- readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising providing the one or more images to a convolutional neural network.
  • the convolutional neural network can computationally analyze the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
  • the one or more non-transitory computer-readable media can also store additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations and generating, based on the group of sequence representations, one or more images that include a plurality of pixels.
  • different forms of DNA are physically partitioned based on one or more characteristics of the DNA. This approach can be used to determine, for example, whether certain sites or regions are hypermethylated or hypomethylated. Partitioning can be performed before attaching adapters to DNA molecules in the sample, e.g., so as to facilitate including partition tags in the adapters. Partition tags can be used to identify which partition a molecule was found in. Following partitioning (and attachment of adapters if applicable), further steps such as amplification, target capture, and sequencing may be performed. [00221] Methylation profiling can involve determining methylation patterns across different regions of the genome.
  • the sequences of molecules in the different partitions can be mapped to a reference genome. This can show regions of the genome that, compared with other regions, are more highly methylated or are less highly methylated. In this way, genomic regions, in contrast to individual molecules, may differ in their extent of methylation.
  • Partitioning nucleic acid molecules in a sample can increase a rare signal, e.g., by enriching rare nucleic acid molecules that are more prevalent in one partition of the sample. For example, a genetic variation present in hypermethylated DNA but less (or not) present in hypomethylated DNA can be more easily detected by partitioning a sample into hypermethylated and hypomethylated nucleic acid molecules. By analyzing multiple partitions of a sample, a multi-dimensional analysis of a single molecule can be performed and hence, greater sensitivity can be achieved. Partitioning may include physically partitioning nucleic acid molecules into partitions or subsamples based on the presence or absence of one or more methylated nucleobases.
  • hypermethylation and/or hypomethylation variable epigenetic target regions are analyzed to determine whether they show differential methylation characteristic of particular immune cell types, such as rare immune cell types, tumor cells or cells of a type that does not normally contribute to the DNA sample being analyzed (such as cfDNA).
  • heterogeneous DNA in a sample is partitioned into two or more partitions (e.g., at least 3, 4, 5, 6 or 7 partitions).
  • each partition is differentially tagged.
  • Tagged partitions can then be pooled together for collective sample prep and/or sequencing.
  • the partitioning-tagging-pooling steps can occur more than once, with each round of partitioning occurring based on a different characteristic (examples provided herein), and tagged using differential tags that are distinguished from other partitions and partitioning means.
  • the differentially tagged partitions are separately sequenced.
  • sequence reads from differentially tagged and pooled DNA are obtained and analyzed in silico.
  • Tags are used to sort reads from different partitions.
  • Analysis to detect genetic variants can be performed on a partition-by-partition level, as well as whole nucleic acid population level.
  • analysis can include in silico analysis to determine genetic variants, such as CNV, SNV, indel, fusion in nucleic acids in each partition.
  • in silico analysis can include determining chromatin structure.
  • coverage of sequence reads can be used to determine nucleosome positioning in chromatin. Higher coverage can correlate with higher nucleosome occupancy in genomic region while lower coverage can correlate with lower nucleosome occupancy or nucleosome depleted region (NDR).
  • partitioning agents include antibodies, such as antibodies that recognize a modified nucleobase, which may be a modified cytosine, such as a methylcytosine (e.g., 5-methylcytosine).
  • the partitioning agent is an antibody that recognizes a modified cytosine other than 5-methylcytosine, such as 5- carboxylcytosine (5caC).
  • Alternative partitioning agents include methyl binding domain (MBDs) and methyl binding proteins (MBPs) as described herein, including proteins such as MeCP2.
  • methylation When using MeDIP or MethylMinerOMethylated DNA Enrichment Kit (ThermoFisher Scientific) various levels of methylation can be partitioned using sequential elutions. For example, a hypomethylated partition (no methylation) can be separated from a methylated partition by contacting the nucleic acid population with the MBD from the kit, which is attached to magnetic beads. The beads are used to separate out the methylated nucleic acids from the non- methylated nucleic acids. Subsequently, one or more elution steps are performed sequentially to elute nucleic acids having different levels of methylation.
  • a first set of methylated nucleic acids can be eluted at a salt concentration of 160 mM or higher, e.g., at least 150 mM, at least 200 mM, 300 mM, 400 mM, 500 mM, 600 mM, 700 mM, 800 mM, 900 mM, 1000 mM, or 2000 mM.
  • a salt concentration 160 mM or higher, e.g., at least 150 mM, at least 200 mM, 300 mM, 400 mM, 500 mM, 600 mM, 700 mM, 800 mM, 900 mM, 1000 mM, or 2000 mM.
  • the partitioning is performed by contacting the nucleic acids with a methyl binding domain (“MBD”) of a methyl binding protein (“MBP”).
  • MBD methyl binding domain
  • MBP methyl binding protein
  • the nucleic acids are contacted with an entire MBP.
  • an MBD binds to 5-methylcytosine (5mC)
  • an MBP comprises an MBD and is referred to interchangeably herein as a methyl binding protein or a methyl binding domain protein.
  • an MBD binds to 5mC and 5hmC.
  • MBD is coupled to paramagnetic beads, such as Dynabeads® M-280 Streptavidin via a biotin linker.
  • MeCP2 is a protein that preferentially binds to 5-methyl-cytosine over unmodified cytosine.
  • RPL26, PRP8 and the DNA mismatch repair protein MHS6 preferentially bind to 5- hydroxymethyl-cytosine over unmodified cytosine.
  • FOXK1 , FOXK2, FOXP1 , FOXP4 and FOXI3 preferably bind to 5-formyl cytosine over unmodified cytosine (lurlaro et al., Genome Biol. 14: R119 (2013)).
  • nonspecifically partitioned molecules are removed using a methylation sensitive nuclease, e.g., a methylation sensitive restriction enzyme (MSRE), digesting/cleaving the DNA where the restriction enzyme (RE) recognition site contains an unmethylated nucleotide but not cleaving the DNA where the restriction enzyme (RE) recognition site contains a methylated nucleotide.
  • a hypomethylated subsample is contacted with a methylation-dependent nuclease, such as a methylation-dependent restriction enzyme, thereby degrading nonspecifically partitioned DNA, e.g., methylated DNA, in the subsample.
  • Degradation of nonspecifically partitioned DNA in one or more partitioned subsamples may improve the performance of methods that rely on accurate partitioning of DNA on the basis of a cytosine modification. For example, such degradation may provide improved sensitivity and/or simplify downstream analyses.
  • partitioning DNA on the basis of a modification, such as methylation then removing nonspecifically partitioned DNA using MDREs and/or MSREs as described herein provides improved efficiency and/or cost over DNA analysis methods comprising procedures that affect a first nucleobase differently from a second nucleobase, such as bisulfite sequencing or bisulfite conversion.
  • a methylation-sensitive nuclease comprises one or more of Aatll, Accll, Acil, Aor13HI, Aor15HI, BspT104l, BssHII, BstUI, CfrI Ol, Clal, Cpol, Eco52l, Haell, Hapll, Hhal, Hin6l, Hpall, HpyCH4IV, Mlul, Mspl, Nael, Notl, Nrul, Nsbl, PmaCI, Psp1406l, Pvul, Sacll, Sall, Smal, and SnaBI. In some embodiments, at least two methylation-sensitive nucleases are used.
  • the methylationsensitive nucleases comprise BstUI and Hpall. In some embodiments, the two methylation-sensitive nucleases comprise Hhal and Accll. In some embodiments, the methylation-sensitive nucleases comprise BstUI, Hpall and Hin6l.
  • the DNA is partitioned, comprising contacting the DNA with an agent that preferentially binds to nucleic acids bearing an epigenetic modification.
  • the nucleic acids are partitioned into at least two subsamples differing in the extent to which the nucleic acids bear the modification from binding to the agents. For example, if the agent has affinity for nucleic acids bearing the modification, nucleic acids overrepresented in the modification (compared with median representation in the population) preferentially bind to the agent, whereas nucleic acids underrepresented for the modification do not bind or are more easily eluted from the agent.
  • the nucleic acids can then be amplified from primers binding to the primer binding sites within the adapters.
  • Tags can be molecules, such as nucleic acids, containing information that indicates a feature of the molecule with which the tag is associated.
  • molecules can bear a sample tag (which distinguishes molecules in one sample from those in a different sample) or a molecular tag/molecular barcode/barcode (which distinguishes different molecules from one another (in both unique and non-unique tagging scenarios).
  • a partition tag which distinguishes molecules in one partition from those in a different partition
  • adapters added to DNA molecules comprise tags.
  • different sets of molecular barcodes, or molecular tags can be used such that the barcodes serve as a molecular tag through their individual sequences and also serve to identify the partition and/or sample to which they correspond based the set of which they are a member.
  • two or more partitions is/are differentially tagged.
  • Tags can be used to label the individual polynucleotide population partitions so as to correlate the tag (or tags) with a specific partition.
  • tags can be used in embodiments that do not employ a partitioning step.
  • a single tag can be used to label a specific partition.
  • multiple different tags can be used to label a specific partition.
  • the set of tags used to label one partition can be readily differentiated for the set of tags used to label other partitions.
  • partition tagging comprises tagging molecules in each partition with a partition tag.
  • the partition tags identify the source partition.
  • the partition tags can serve as identifiers of the source partition and the molecule, i.e. , different partitions are tagged with different sets of molecular tags, e.g., comprised of a pair of barcodes.
  • the one or more molecular barcodes attached to the molecule indicates the source partition as well as being useful to distinguish molecules within a partition. For example, a first set of 35 barcodes can be used to tag molecules in a first partition, while a second set of 35 barcodes can be used tag molecules in a second partition.
  • enrichment or capture is performed after attachment of adapters to sample molecules. In some embodiments, enrichment or capture is performed after a partitioning step. In some embodiments, enrichment or capture is performed after an amplification step. In some embodiments, sample molecules are partitioned, then adapters are attached, then sample molecules are amplified, and then the amplified molecules are subjected to enrichment or capture. The enriched or captured molecules may then be subjected to another amplification and then sequenced. [00266] In some embodiments, the probes specific for the target regions comprise a capture moiety that facilitates the enrichment or capture of the DNA hybridized to the probes. In some embodiments, the capture moiety is biotin.
  • captured DNA is amplified while attached to the solid support.
  • the amplification comprises use of a PCR primer that anneals to a sequence within an adapter and a PCR primer that anneals to a sequence within a probe annealed to the target region of the DNA.
  • the capturing step may be performed using conditions suitable for specific nucleic acid hybridization, which generally depend to some extent on features of the probes such as length, base composition, etc. Those skilled in the art will be familiar with appropriate conditions given general knowledge in the art regarding nucleic acid hybridization. In some embodiments, complexes of target-specific probes and DNA are formed.
  • methods described herein comprise capturing a plurality of sets of target regions of cfDNA obtained from a subject.
  • the target regions may comprise differences depending on whether they originated from a tumor or from healthy cells or from a certain cell type.
  • the capturing step produces a captured set of cfDNA molecules.
  • cfDNA molecules corresponding to a sequence-variable target region set are captured at a greater capture yield in the captured set of cfDNA molecules than cfDNA molecules corresponding to an epigenetic target region set.
  • a method described herein comprises contacting cfDNA obtained from a subject with a set of target-specific probes, wherein the set of target-specific probes is configured to capture cfDNA corresponding to the sequencevariable target region set at a greater capture yield than cfDNA corresponding to the epigenetic target region set.
  • the volume of data needed to determine fragmentation patterns (e.g., to test for perturbation of transcription start sites or CTCF binding sites) or fragment abundance (e.g., in hypermethylated and hypomethylated partitions) is generally less than the volume of data needed to determine the presence or absence of cancer-related sequence mutations.
  • Capturing the target region sets at different yields can facilitate sequencing the target regions to different depths of sequencing in the same sequencing run (e.g., using a pooled mixture and/or in the same sequencing cell).
  • a capturing step is performed with probes for a sequence-variable target region set and probes for an epigenetic target region set in the same vessel at the same time, e.g., the probes for the sequence-variable and epigenetic target region sets are in the same composition.
  • concentration of the probes for the sequence-variable target region set is greater that the concentration of the probes for the epigenetic target region set.
  • the second nucleobase is a modified or unmodified adenine; if the first nucleobase is a modified or unmodified cytosine, then the second nucleobase is a modified or unmodified cytosine; if the first nucleobase is a modified or unmodified guanine, then the second nucleobase is a modified or unmodified guanine; and if the first nucleobase is a modified or unmodified thymine, then the second nucleobase is a modified or unmodified thymine (where modified and unmodified uracil are encompassed within modified thymine for the purpose of this step).
  • the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises bisulfite conversion.
  • Treatment with bisulfite converts unmodified cytosine and certain modified cytosine nucleotides (e.g. 5-formyl cytosine (fC) or 5-carboxylcytosine (caC)) to uracil whereas other modified cytosines (e.g., 5-methylcytosine, 5-hydroxylmethylcystosine) are not converted.
  • modified cytosine nucleotides e.g. 5-formyl cytosine (fC) or 5-carboxylcytosine (caC)
  • fC 5-formyl cytosine
  • caC 5-carboxylcytosine
  • the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises chemical-assisted conversion with a substituted borane reducing agent, optionally wherein the substituted borane reducing agent is 2-picoline borane, borane pyridine, tert-butylamine borane, or ammonia borane.
  • a substituted borane reducing agent such as potassium perruthenate (KRuO4) (also suitable for use in ox-BS conversion) is used to specifically oxidize hmC to fC.
  • the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises APOBEC-coupled epigenetic (ACE) conversion.
  • ACE conversion an AID/APOBEC family DNA deaminase enzyme such as APOBEC3A (A3A) is used to deaminate unmodified cytosine and mC without deaminating hmC, fC, or caC.
  • A3A APOBEC3A
  • the first nucleobase comprises unmodified C and/or mC (e.g., unmodified C and optionally mC)
  • the second nucleobase comprises hmC.
  • Sequencing of ACE-converted DNA identifies positions that are read as cytosine as being hmC, fC, or caC positions. Meanwhile, positions that are read as T are identified as being T, unmodified C, or mC. Performing ACE conversion on a DNA sample as described herein thus facilitates distinguishing positions containing hmC from positions containing mC or unmodified C using the sequence reads obtained from the sample.
  • ACE conversion see, e.g., Schutsky et al., Nature Biotechnology 2018; 36: 1083-1090.
  • the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises enzymatic conversion of the first nucleobase, e.g., as in EM-Seq. See, e.g., Vaisvila R, et al. (2019) EM-seq: Detection of DNA methylation at single base resolution from picograms of DNA. bioRxiv; DOI: 10.1101/2019.12.20.884692, available at www. biorxiv.org/content/10.1101/2019.12.20.884692v1 .
  • TET2 and T4-[3GT or 5-hydroxymethylcytosine carbamoyltransferase can be used to convert 5mC and 5hmC into substrates that cannot be deaminated by a deaminase (e.g., APOBEC3A), and then a deaminase (e.g., APOBEC3A) can be used to deaminate unmodified cytosines converting them to uracils.
  • the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises enzymatic conversion of the first nucleobase using a non-specific, modification-sensitive double-stranded DNA deaminase, e.g., as in SEM-seq.
  • a non-specific, modification-sensitive double-stranded DNA deaminase e.g., as in SEM-seq.
  • SEM-Seq employs a non-specific, modification-sensitive double-stranded DNA deaminase (MsddA) in a nondestructive single-enzyme 5-methylctyosine sequencing (SEM-seq) method that deaminates unmodified cytosines.
  • MsddA modification-sensitive double-stranded DNA deaminase
  • SEM-seq nondestructive single-enzyme 5-methylctyosine sequencing
  • Cytosines that are read as thymines are identified as unmodified (e.g., unmethylated) cytosines or as thymines in the DNA. Performing SEM-seq conversion thus facilitates identifying positions containing 5mC using the sequence reads obtained.
  • the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises enzymatic conversion of the first nucleobase using MsddA.
  • the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA of the first subsample converts a modified nucleoside.
  • the conversion procedure which converts a modified nucleosides comprises enzymatic conversion, such as DM-seq, for example, as described in WO2023/288222A1 .
  • DM-seq unmodified cytosines in the DNA are enzymatically protected from a subsequent deamination step wherein 5mC in 5mCpG is converted to T.
  • the enzymatically protected unmodified (e.g., unmethylated) cytosines are not converted and are read as “C” during sequencing.
  • Cytosines that are read as thymines are identified as methylated cytosines in the DNA.
  • the first nucleobase comprises unmodified (such as unmethylated) cytosine
  • the second nucleobase comprises modified (such as methylated) cytosine.
  • Sequencing of the converted DNA identifies positions that are read as cytosine as being unmodified C positions. Meanwhile, positions that are read as T are identified as being T or 5mC. Performing DM-seq conversion thus facilitates identifying positions containing 5mC using the sequence reads obtained.
  • methyltransferase is used broadly herein to refer to enzymes capable of transferring a methyl or substituted methyl (e.g., carboxymethyl) to a substrate (e.g., a cytosine in a nucleic acid).
  • a substrate e.g., a cytosine in a nucleic acid.
  • the DNA is contacted with a CpG-specific DNA methyltransferase (MTase), such as a CpG-specific carboxym ethyltransferase (CxMTase), and a substituted methyl donor, such as a carboxymethyl donor (e.g., carboxymethyl-S-adenosyl-L-methionine).
  • MTase CpG-specific DNA methyltransferase
  • CxMTase CpG-specific carboxym ethyltransferase
  • a substituted methyl donor such as a carboxymethyl donor
  • the CxMTase can facilitate the addition of a protective carboxymethyl group to an unmethylated cytosine.
  • the unmethylated cytosine is unmodified cytosine.
  • the carboxymethyl group can prevent deamination of the cytosine during a deamination step (such as a deamination step using an APOBEC enzyme, such as A3A).
  • Substituted methyl or carboxymethyl donors useful in the disclosed methods include but are not limited to, S-adenosyl-L-methionine (SAM) analogs, optionally wherein the SAM analog is carboxy-S-adenosyl-L-methionine (CxSAM).
  • the MTase may be, for example, a CpG methyltransferase from Spiroplasma sp. strain MQ1 (M.Sssl), DNA-methyltransferase 1 (DNMT1 ), DNA-methyltransferase 3 alpha (DNMT3A), DNA-methyltransferase 3 beta (DNMT3B), or DNA adenine methyltransferase (Dam).
  • the CxMTase may be a CpG methyltransferase from Mycoplasma penetrans (M.Mpel).
  • the methyltransferase enzyme is a variant of M.Mpel, wherein the amino acid corresponding to position 374 is R or K.
  • the methyltransferase enzyme is a variant of M.Mpel having an N374R substitution or an N374K substitution.
  • the methyltransferase variant can further comprise one or more amino acid substitutions selected from a) substitution of one or both residues T300 and E305 with S, A, G, Q, D, or N; b) substitution of one or more residues A323, N306, and Y299 with a positively charged amino acid selected from K, R or H; and/or c) substitution of S323 with A, G, K, R or H, which may enhance the activity of the enzyme.
  • the conversion procedure further includes enzymatic protection of 5hmCs, such as by glucosylation of the 5hmCs (e.g., using [3GT) or by carbamoylation of the 5hmCs (e.g., using 5-hydroxymethylcytosine carbamoyltransferase), in the DNA prior to the deamination of unprotected modified cytosines.
  • enzymatic protection of 5hmCs such as by glucosylation of the 5hmCs (e.g., using [3GT) or by carbamoylation of the 5hmCs (e.g., using 5-hydroxymethylcytosine carbamoyltransferase), in the DNA prior to the deamination of unprotected modified cytosines.
  • 5hmC can be protected from conversion, for example through glucosylation using [3-glucosyl transferase (PGT), forming (5-glucosylhydroxymethylcytosine) 5ghmC, or through carbamoylation using 5-hydroxymethylcytosine carbamoyltransferase, forming 5cmC.
  • PTT [3-glucosyl transferase
  • 5cmC 5-hydroxymethylcytosine carbamoyltransferase
  • Glucosylation or carbamoylation of 5hmC can reduce or eliminate deamination of 5hmC by a deaminase such as APOBEC3A.
  • Treatment with an MTase or CxMTase then adds a protecting group to unmodified (unmethylated) cytosines in the DNA.
  • 5mC (but not protected, unmodified cytosine and not 5ghmC or 5cmC) is then deaminated (converted to T in the case of 5mC) by treatment with a deaminase, for example, an APOBEC enzyme (such as APOBEC3A).
  • a deaminase for example, an APOBEC enzyme (such as APOBEC3A).
  • Sequencing of the converted DNA identifies positions that are read as cytosine as being either 5hmC or unmodified C positions. Meanwhile, positions that are read as T are identified as being T or 5mC. Performing DM-seq conversion with glucosylation of 5hmC on a sample as described herein thus facilitates distinguishing positions containing unmodified C or 5hmC on the one hand from positions containing 5
  • cytosines can be left intact while methylated cytosines and hydroxymethylcytosines are converted to a base read as a thymine (e.g., uracil, thymine, or dihydrouracil).
  • a thymine e.g., uracil, thymine, or dihydrouracil
  • methylating a cytosine in at least one first complementary strand or second complementary strand comprises contacting the cytosine with a methyltransferase such as DNMT1 or DNMT5.
  • a methyltransferase such as DNMT1 or DNMT5.
  • the step of oxidizing a 5-hydroxymethylated cytosine to 5-formylcytosine can be optional.
  • converting the modified cytosine in at least one first or second strand to a thymine or a base read as thymine comprises oxidizing a hydroxymethyl cytosine, e.g., the hydroxymethyl cytosine is oxidized to formylcytosine.
  • oxidizing the hydroxymethyl cytosine to formylcytosine comprises contacting the hydroxymethyl cytosine with a ruthenate, such as potassium ruthenate (KRuO4).
  • converting the formylcytosine and/or the methylcytosine to carboxylcytosine can comprise contacting the formylcytosine and/or the methylcytosine with a TET enzyme, such as TET1 , TET2, TET3, or a TET2 comprising a T1372S mutation.
  • the method comprises reducing the carboxylcytosine as part of converting the modified cytosine in at least one first or second strand to a thymine or a base read as thymine, and/or the carboxylcytosine is reduced to dihydrouracil.
  • reducing the carboxylcytosine comprises contacting the carboxylcytosine with a borane or borohydride reducing agent.
  • the borane or borohydride reducing agent comprises pyridine borane, 2-picoline borane, borane, tert-butylamine borane, ammonia borane, sodium borohydride, sodium cyanoborohydride (NaBH3CN), lithium borohydride (LiBH4), ethylenediamine borane, dimethylamine borane, sodium triacetoxyborohydride, morpholine borane, 4-methylmorpholine borane, trimethylamine borane, dicyclohexylamine borane, or a salt thereof.
  • the one or more TET enzymes may be used in the disclosed methods as appropriate.
  • the one or more TET enzymes comprise TETv.
  • TETv is described in US Patent 10,260,088.
  • the one or more TET enzymes comprise TETcd.
  • TETcd is described in US Patent 10,260,088.
  • the one or more TET enzymes comprise TET 1 .
  • the one or more TET enzymes comprise TET2.
  • TET2 may be expressed and used as a fragment comprising TET2 residues 1129-1480 joined to TET2 residues 1844-1936 by a linker as described, e.g., in US Patent 10,961 ,525.
  • the one or more TET enzymes comprise TET1 and TET2. In some embodiments, the one or more TET enzymes comprise a V1900 TET mutant, such as a V1900A, V1900C, V1900G, V1900I, or V1900P TET mutant. In some embodiments, the one or more TET enzymes comprise a V1900 TET2 mutant, such as a V1900A, V1900C, V1900G, V1900I, or V1900P TET2 mutant.
  • the TET enzyme can be beneficial to use a TET enzyme that maximizes formation of 5-carboxylcytosine (5-caC) relative to less oxidized modified cytosines, particularly 5- formylcytosine, because 5-caC is not a substrate for enzymatic deamination, e.g., by APOBEC enzymes such as APOBEC3A. Maximizing formation of 5-caC thus reduces the risk of false calls in which a base is identified as unmethylated because it underwent deamination even though it was methylated (or hydroxymethylated) in the original sample. Accordingly, in some embodiments, the TET enzyme comprises a mutation that increases formation of 5-caC.
  • the one or more TET enzymes comprise a TET2 enzyme comprising a T 1372S mutation, such as TET2-CS-T 1372S and TET2-CD- T1372S.
  • TET2 comprising a T1372S mutation is described in US Patent 10,961 ,525 and may be expressed and used as a fragment comprising TET2 residues 1129-1480 joined to TET2 residues 1844-1936 by a linker.
  • Position 1372 of TET2 corresponds to position 258 of SEQ ID NO: 21 (wild type TET2 catalytic domain) of US Patent 10,961 ,525.
  • the sequence of a T1372S TET2 catalytic domain may be obtained by changing the threonine at position 258 of SEQ ID NO: 21 of US Patent 10,961 ,525 to serine.
  • TET2 comprising a T1372S mutation is also described in Liu et al., Nat Chem Biol. 2017 February; 13(2): 181-187. As demonstrated in Liu et al., TET2 comprising a T1372S mutation can more efficiently oxidize 5mC to produce 5-carboxylcytosine (5caC) than other versions of TET2 such as TET2 lacking a T1372S mutation.
  • the TET2 enzyme is a human TET2 enzyme comprising a T1372S mutation.
  • a mutation that increases formation of 5-caC means that the TET enzyme having the mutation produces more 5-caC than a TET enzyme that lacks the mutation but is otherwise identical.
  • 5-caC production can be measured as described, e.g., in Liu et al., Nat Chem Biol 13:181 -187 (2017) (see Online Methods section, TET reactions in vitro subsection, “driving” conditions). Any variants and/or mutants described in Liu et al. (2017) can be used in the disclosed methods as appropriate.
  • the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises separating DNA originally comprising the first nucleobase from DNA not originally comprising the first nucleobase.
  • the first nucleobase is hmC.
  • DNA originally comprising the first nucleobase may be separated from other DNA using a labeling procedure comprising biotinylating positions that originally comprised the first nucleobase.
  • the first nucleobase is first derivatized with an azide- containing moiety, such as a glucosyl-azide containing moiety.
  • the azide-containing moiety then may serve as a reagent for attaching biotin, e.g., through Huisgen cycloaddition chemistry.
  • biotin-binding agent such as avidin, neutravidin (deglycosylated avidin with an isoelectric point of about 6.3), or streptavidin.
  • hmC-seal An example of a procedure for separating DNA originally comprising the first nucleobase from DNA not originally comprising the first nucleobase is hmC-seal, which labels hmC to form [3-6-azide-glucosyl-5- hydroxymethylcytosine and then attaches a biotin moiety through Huisgen cycloaddition, followed by separation of the biotinylated DNA from other DNA using a biotin-binding agent.
  • hmC-seal see, e.g., Han et al., Mol. Cell 2016; 63: 711 -719. This approach is useful for identifying fragments that include one or more hmC nucleobases.
  • the method further comprises differentially tagging each of the DNA originally comprising the first nucleobase, the DNA not originally comprising the first nucleobase.
  • the method may further comprise pooling the DNA originally comprising the first nucleobase and the DNA not originally comprising the first nucleobase following differential tagging.
  • the DNA originally comprising the first nucleobase and the DNA not originally comprising the first nucleobase may then be used in downstream analyses.
  • the pooled DNA originally comprising the first nucleobase and the DNA not originally comprising the first nucleobase may be sequenced in the same sequencing cell (such as after being subjected to further treatments, such as those described herein) while retaining the ability to resolve whether a given read came from a molecule of DNA originally comprising the first nucleobase or DNA not originally comprising the first nucleobase using the differential tags.
  • the first nucleobase is a modified or unmodified adenine
  • the second nucleobase is a modified or unmodified adenine.
  • the modified adenine is N6-methyladenine (mA).
  • the modified adenine is one or more of N6-methyladenine (mA), N6- hydroxymethyladenine (hmA), or N6-formyladenine (fA).
  • Techniques comprising partitioning based on methylation status or methylated DNA immunoprecipitation (MeDIP) can be used to separate DNA containing modified bases such as mC, mA, caC (which may be generated by oxidation of mC or hmC with Tet2, e.g., before enzymatic conversion of unmodified C to II, e.g., using a deaminase such as APOBEC3A), or dihydrouracil from other DNA.
  • modified bases such as mC, mA, caC (which may be generated by oxidation of mC or hmC with Tet2, e.g., before enzymatic conversion of unmodified C to II, e.g., using a deaminase such as APOBEC3A), or dihydrouracil from other DNA.
  • mA An antibody specific for mA is described in Sun et al., Bioessays 2015; 37:1155-62.
  • Antibodies for various modified nucleobases such as mC, caC, and forms of thymine/uracil including dihydrouracil or halogenated forms such as 5-bromouracil, are commercially available.
  • Various modified bases can also be detected based on alterations in their base pairing specificity.
  • hypoxanthine is a modified form of adenine that can result from deamination and is read in sequencing as a G. See, e.g., US Patent 8,486,630; Brown, Genomes, 2nd Ed., John Wiley & Sons, Inc., New York, N.Y., 2002, chapter 14, “Mutation, Repair, and Recombination.”
  • nucleic acids captured or enriched using a method described herein comprise captured DNA, such as one or more captured sets of DNA.
  • the captured DNA comprise target regions that are differentially methylated in different immune cell types.
  • the immune cell types comprise rare or closely related immune cell types, such as activated and naive lymphocytes or myeloid cells at different stages of differentiation.
  • a captured epigenetic target region set captured from a sample or a subsample comprises genomic regions that show no or negligible methylation signal when analyzing cell-free DNA (cfDNA) from healthy individuals (e.g. in blood) but exhibit detectable methylation when analyzing cfDNA from individuals with cancer.
  • cfDNA cell-free DNA
  • Such regions are characterized by low background methylation levels in healthy populations, thereby providing an enhanced contrast that facilitates sensitive detection of tumor-derived DNA.
  • hypomethylation target regions may be hypomethylated in other cell types but not to the extent observed in the one cell type.
  • the hypomethylation target regions show higher methylation in healthy cfDNA than in at least one other tissue type.
  • proliferating or activated immune cells may shed more DNA into the bloodstream than immune cells in a healthy individual (and healthy cells of the same tissue type, respectively).
  • the distribution of cell type and/or tissue of origin of cfDNA may change upon carcinogenesis.
  • the distribution of immune cell type of origin may change in a subject having cancer, precancer, infection, transplant rejection, or other disease or disorder directly or indirectly affecting the immune system.
  • the status of epigenetic target regions of certain immune cell types likewise may change in a subject having such a disease relative to a healthy subject or relative to the same subject prior to having the disease or disorder.
  • variations in hypermethylation and/or hypomethylation can be an indicator of disease.
  • an increase in the level of hypermethylation target regions and/or hypomethylation target regions in a subsample following a partitioning step can be an indicator of the presence (or recurrence, depending on the history of the subject) of cancer.
  • Exemplary hypermethylation target regions and hypomethylation target regions useful for distinguishing between various cell types have been identified by analyzing DNA obtained from various cell types via whole genome bisulfite sequencing, as described, e.g., in Stunnenberg, H. G. et. al., “The International Human Epigenome Consortium: A Blueprint for Scientific Collaboration and Discovery,” Cell 167, 1145 (2016) (doi.org/10.1186/sl3059-020-02065- 5).
  • Whole-genome bisulfite sequencing data is available from the Blueprint consortium, available on the internet at dcc.blueprint-epigenome.eu.
  • first and second captured target region sets comprise, respectively, DNA corresponding to a sequence-variable target region set and DNA corresponding to an epigenetic target region set, for example, as described in WO 2020/160414.
  • the first and second captured sets may be combined to provide a combined captured set.
  • DNA e.g., a sample or subsample
  • enrichment or capture may use oligonucleotides (e.g., primers or probes) specific for the altered or unaltered sequence, as desired.
  • the DNA corresponding to the sequence-variable target region set may be present at a greater concentration than the DNA corresponding to the epigenetic target region set, e.g., a 1.1 to 1.2-fold greater concentration, a 1.2- to 1.4-fold greater concentration, a 1.4- to 1.6-fold greater concentration, a 1.6- to 1.8-fold greater concentration, a 1.8- to 2.0-fold greater concentration, a 2.0- to 2.2-fold greater concentration, a 2.2- to 2.4-fold greater concentration a 2.4- to 2.6-fold greater concentration, a 2.6- to 2.8-fold greater concentration, a 2.8- to 3.0-fold greater concentration, a 3.0- to 3.5-fold greater concentration, a 3.5- to 4.0, a 4.0- to 4.5-fold greater concentration, a 4.5- to 5.0
  • an epigenetic target region set may comprise one or more types of target regions likely to differentiate DNA from different immune cell types and other non- immune cell types and/or to differentiate neoplastic (e.g., tumor or cancer) cells and from healthy cells, e.g., non-neoplastic circulating cells. Exemplary types of such regions are discussed in detail herein.
  • the epigenetic target region set may also comprise one or more control regions, e.g., as described herein.
  • the epigenetic target region set has a footprint of at least 100 kb, e.g., at least 200 kb, at least 300 kb, or at least 400 kb. In some embodiments, the epigenetic target region set has a footprint in the range of 100-1000 kb, e.g., 100-200 kb, 200-300 kb, 300- 400 kb, 400-500 kb, 500-600 kb, 600-700 kb, 700- 800 kb, 800-900 kb, and 900-1 ,000 kb.
  • Methylation patterns of hypermethylation target regions that are useful for deconvoluting immune cell types may further change in certain disease states, such as cancer.
  • hypermethylation target regions that are useful for deconvoluting immune cell types are also useful for determining the likelihood that the subject from which the sample was obtained has cancer or precancer.
  • hypermethylation target regions are useful for determining whether levels of particular immune cell types are abnormal and whether such abnormal levels are likely related to the presence of cancer or precancer, or if they are related to a different disease or condition other than cancer or precancer.
  • hypermethylation target regions can include regions that do not necessarily differ in methylation in cancerous tissue relative to DNA from healthy tissue of the same type, but do differ in methylation (e.g., have more methylation) relative to cfDNA that is typical in healthy subjects.
  • a cancer results in increased cell death such as apoptosis of cells of the tissue type corresponding to the cancer, such a cancer can be detected at least in part using such hypermethylation target regions.
  • hypermethylation target regions useful for determining the likelihood that a subject has cancer are different than the hypermethylation target regions useful for determining the levels of particular immune cell types. In some embodiments, at least some of the hypermethylation target regions useful for determining the likelihood that a subject has cancer are the same as the hypermethylation target regions useful for determining the levels of particular immune cell types.
  • Methylation variable target regions in various types of lung cancer are discussed in detail, e.g., in Ooki et al., Clin. Cancer Res. 23:7141 -52 (2017); Belinksy, Annu. Rev. Physiol. 77:453- 74 (2015); Hulbert et al., Clin. Cancer Res. 23:1998-2005 (2017); Shi et al., BMC Genomics 18:901 (2017); Schneider et al., BMC Cancer. 11 :102 (2011 ); Lissa et al., Transl Lung Cancer Res 5(5):492-504 (2016); Skvortsova et al., Br. J. Cancer.
  • the hypermethylation target regions comprise a plurality of loci listed in Table 1 or Table 2, e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the loci listed in Table 1 or Table 2.
  • the hypermethylation target regions comprise regions of one or more genes listed in Table 2, e.g.
  • the hypermethylation target regions comprise regions of a plurality of genes listed in Table 2, e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the genes listed in Table 3. In some embodiments, the hypermethylation target regions comprise regions of all of the genes listed in Table 3.
  • Table 3 Exemplary genes comprising exemplary hypermethylation target regions.
  • Exemplary specific genomic regions that show cancer-associated hypomethylation include nucleotides 8403565-8953708 and 151104701-151106035 of human chromosome 1.
  • the hypomethylation target regions overlap or comprise one or both of these regions.
  • the epigenetic target regions captured from the second subsample comprise hypomethylation target regions.
  • the epigenetic target regions captured from the second subsample comprise hypomethylation target regions and the epigenetic target regions captured from the first subsample comprise hypermethylation target regions.
  • CTCF is a DNA-binding protein that contributes to chromatin organization and often colocalizes with cohesin. Perturbation of CTCF binding sites has been reported in a variety of different cancers. See, e.g., Katainen et al., Nature Genetics, doi:10.1038/ng.3335, published online 8 June 2015; Guo et al., Nat. Commun. 9:1520 (2018). CTCF binding results in recognizable patterns in cfDNA that can be detected by sequencing, e.g., through fragment length analysis. Details regarding sequencing-based fragment length analysis are provided in Snyder et al., Cell 164:57-68 (2016); WO 2018/009723; and US20170211143A1 , each of which are incorporated herein by reference.
  • CTCF binding sites are a type of fragmentation variable target regions.
  • CTCF binding sites there are many known CTCF binding sites. See, e.g., the CTCFBSDB (CTCF Binding Site Database), available on the Internet at insulatordb.uthsc.edu/; Cuddapah et al., Genome Res. 19:24-32 (2009); Martin et al., Nat. Struct. Mol. Biol. 18:708-14 (2011 ); Rhee et al., Cell. 147:1408-19 (2011 ), each of which are incorporated by reference.
  • Exemplary CTCF binding sites are at nucleotides 56014955-56016161 on chromosome 8 and nucleotides 95359169-95360473 on chromosome 13.
  • the epigenetic target region set includes CTCF binding regions.
  • the CTCF binding regions comprise at least 10, 20, 50, 100, 200, or 500 CTCF binding regions, or 10-20, 20-50, 50- 100, 100-200, 200-500, or 500-1000 CTCF binding regions, e.g., such as CTCF binding regions described above or in one or more of CTCFBSDB or the Cuddapah et al., Martin et al., or Rhee et al. articles cited above.
  • the CTCF sites can be methylated or unmethylated, wherein the methylation state is correlated with the whether or not the cell is a cancer cell.
  • the epigenetic target region set comprises at least 100 bp, at least 200 bp, at least 300 bp, at least 400 bp, at least 500 bp, at least 750 bp, at least 1000 bp upstream and downstream regions of the CTCF binding sites. d. Transcription start sites.
  • T ranscription start sites may also show perturbations in neoplastic cells.
  • nucleosome organization at various transcription start sites in healthy cells of the hematopoietic lineage — which contributes substantially to cfDNA in healthy individuals — may differ from nucleosome organization at those transcription start sites in neoplastic cells. This results in different cfDNA patterns that can be detected by sequencing, as discussed generally in Snyder et al., Cell 164:57-68 (2016); WO 2018/009723; and US20170211143A1.
  • the DNA is obtained from a subject having neoplasia. In some embodiments, the DNA (e.g., cfDNA) is obtained from a subject suspected of having neoplasia. In some embodiments, the DNA (e.g., cfDNA) is obtained from a subject in remission from a tumor, cancer, or neoplasia (e.g., following chemotherapy, surgical resection, radiation, or a combination thereof). In any of the foregoing embodiments, the cancer, tumor, or neoplasia or suspected cancer, tumor, or neoplasia may be of the lung, colon, rectum, kidney, breast, prostate, or liver.
  • sequencing comprises detecting and/or distinguishing unmodified and modified nucleobases.
  • PacBio sequencing e.g., single-molecule real-time (SMRT) sequencing
  • SMRT single-molecule real-time
  • Oxford nanopore sequencing systems e.g., MinlON sequencer
  • methylation of DNA for example: 5-methylcytosine and 5-hydroxymethylcytosine
  • cell-free nucleic acids may be sequenced with at least 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 50000, 100,000 sequencing reactions. In other cases, cell-free nucleic acids may be sequenced with less than 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 50000, 100,000 sequencing reactions. Sequencing reactions may be performed sequentially or simultaneously. Subsequent data analysis may be performed on all or part of the sequencing reactions. In some cases, data analysis may be performed on at least 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 50000, 100,000 sequencing reactions.
  • said depth of sequencing is at least 2-fold greater. In some embodiments, said depth of sequencing is at least 5-fold greater. In some embodiments, said depth of sequencing is at least 10-fold greater. In some embodiments, said depth of sequencing is 4- to 10-fold greater. In some embodiments, said depth of sequencing is 4- to 100-fold greater.
  • the present methods can be used to diagnose presence of conditions, particularly cancer or precancer, in a subject, to characterize conditions (e.g., staging cancer or determining heterogeneity of a cancer), monitor response to treatment of a condition, effect prognosis risk of developing a condition or subsequent course of a condition.
  • the present disclosure can also be useful in determining the efficacy of a particular treatment option. For example, the change in the tumor fraction or determining the methylation status of one or regions can be useful in determining whether the patient is responding to the treatment or not.
  • certain treatment options may be correlated with methylation profiles of cancers over time. This correlation may be useful in selecting a therapy.
  • the types and number of cancers that may be detected may include blood cancers, brain cancers, lung cancers, skin cancers, nose cancers, throat cancers, liver cancers, bone cancers, lymphomas, pancreatic cancers, skin cancers, bowel cancers, rectal cancers, thyroid cancers, bladder cancers, kidney cancers, mouth cancers, stomach cancers, solid state tumors, heterogeneous tumors, homogenous tumors and the like.
  • Type and/or stage of cancer can be detected from genetic variations including mutations, rare mutations, indels, copy number variations, transversions, translocations, recombination, inversion, deletions, aneuploidy, partial aneuploidy, polyploidy, chromosomal instability, chromosomal structure alterations, gene fusions, chromosome fusions, gene truncations, gene amplification, gene duplications, chromosomal lesions, DNA lesions, abnormal changes in nucleic acid chemical modifications, abnormal changes in epigenetic patterns, and abnormal changes in nucleic acid 5-methylcytosine.
  • Genetic data can also be used for characterizing a specific form of cancer. Cancers are often heterogeneous in both composition and staging.
  • an abnormal condition is cancer or precancer.
  • the abnormal condition may be one resulting in a heterogeneous genomic population.
  • some tumors are known to comprise tumor cells in different stages of the cancer.
  • heterogeneity may comprise multiple foci of disease. Again, in the example of cancer, there may be multiple tumor foci, perhaps where one or more foci are the result of metastases that have spread from a primary site.
  • the present methods can be used to generate or profile, fingerprint or set of data that is a summation of genetic information derived from different cells in a heterogeneous disease.
  • This set of data may comprise copy number variation, epigenetic variation, or other mutation analyses alone or in combination.
  • the present methods can be used to diagnose, prognose, monitor or observe cancers, or other diseases.
  • the methods herein do not involve the diagnosing, prognosing or monitoring a fetus and as such are not directed to non-invasive prenatal testing.
  • these methodologies may be employed in a pregnant subject to diagnose, prognose, monitor or observe cancers or other diseases in an unborn subject whose DNA and other polynucleotides may cocirculate with maternal molecules.
  • An exemplary method for determining an indication of cancer through NGS comprises the following steps:
  • Another exemplary method for determining an indication of cancer through NGS comprises the following steps:
  • Another exemplary method for determining methylation status of a target region (e.g., promoter region) through NGS comprises the following steps:
  • Another exemplary method for determining an indication of cancer or for determining methylation status of a target region (e.g., promoter region) through NGS comprises the following steps:
  • the exemplary methods discussed above can also be used with DNA samples obtained from tissue sample, stool sample or bodily fluids like urine sample.
  • the DNA can be a whole genomic DNA.
  • an additional step of fragmenting the DNA is performed to the methods discussed above.
  • the molecular tags do not comprise nucleotides that are altered by a procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA, such as any of those described herein (e.g., the tags do not comprise unmodified C where the procedure is bisulfite conversion or any other conversion that affects C; the tags do not comprise mC where the procedure is a conversion that affects mC; the tags do not comprise hmC where the procedure is a conversion that affects hmC; etc.).
  • a sample can be any biological sample isolated from a subject.
  • a sample can be a bodily sample.
  • Samples can include body tissues, such as known or suspected solid tumors, whole blood, platelets, serum, plasma, stool, red blood cells, white blood cells or leucocytes, endothelial cells, tissue biopsies, cerebrospinal fluid synovial fluid, lymphatic fluid, ascites fluid, interstitial or extracellular fluid, the fluid in spaces between cells, including gingival crevicular fluid, bone marrow, pleural effusions, cerebrospinal fluid, saliva, mucous, sputum, semen, sweat, urine. Samples are preferably body fluids, particularly blood and fractions thereof, and urine.
  • a sample can be in the form originally isolated from a subject or can have been subjected to further processing to remove or add components, such as cells, or enrich for one component relative to another.
  • a preferred body fluid for analysis is plasma or serum containing cell-free nucleic acids.
  • a population of nucleic acids is obtained from a serum, plasma or blood sample from a subject suspected of having neoplasia, a tumor, precancer, or cancer or previously diagnosed with neoplasia, a tumor, precancer, or cancer.
  • the population includes nucleic acids having varying levels of sequence variation, epigenetic variation, and/or post replication or transcriptional modifications. Postreplication modifications include modifications of cytosine, particularly at the 5-position of the nucleobase, e.g., 5-methylcytosine, 5- hydroxymethylcytosine, 5-formylcytosine and 5-carboxylcytosine.
  • a sample can be isolated or obtained from a subject and transported to a site of sample analysis.
  • the sample may be preserved and shipped at a desirable temperature, e.g., room temperature, 4°C, -20°C, and/or -80°C.
  • a sample can be isolated or obtained from a subject at the site of the sample analysis.
  • the subject can be a human, a mammal, an animal, a companion animal, a service animal, or a pet.
  • the subject may have a cancer, precancer, infection, transplant rejection, or other disease or disorder related to changes in the immune system.
  • the subject may not have cancer or a detectable cancer symptom.
  • the subject may have been treated with one or more cancer therapy, e.g., any one or more of chemotherapies, antibodies, vaccines or biologies.
  • the subject may be in remission.
  • the subject may or may not be diagnosed of being susceptible to cancer or any cancer-associated genetic mutations/disorders.
  • the sample comprises plasma.
  • the volume of plasma obtained can depend on the desired read depth for sequenced regions. Exemplary volumes are 0.4-40 ml, 5-20 ml, 10-20 ml. For examples, the volume can be 0.5 mL, 1 mL, 5 mL 10 mL, 20 mL, 30 mL, or 40 mL. A volume of sampled plasma may be 5 to 20 mL.
  • a sample can comprise various amount of nucleic acid that contains genome equivalents.
  • a sample of about 30 ng DNA can contain about 10,000 (10 4 ) haploid human genome equivalents and, in the case of cfDNA, about 200 billion (2xlO n ) individual polynucleotide molecules.
  • a sample of about 100 ng of DNA can contain about 30,000 haploid human genome equivalents and, in the case of cfDNA, about 600 billion individual molecules.
  • a sample can comprise nucleic acids from different sources, e.g., from cells and cell-free of the same subject, from cells and cell-free of different subjects.
  • a sample can comprise nucleic acids carrying mutations.
  • a sample can comprise DNA carrying germline mutations and/or somatic mutations.
  • Germline mutations refer to mutations existing in germline DNA of a subject.
  • Somatic mutations refer to mutations originating in somatic cells of a subject, e.g., precancer cells or cancer cells.
  • a sample can comprise DNA carrying cancer-associated mutations (e.g., cancer-associated somatic mutations).
  • a sample can comprise an epigenetic variant (i.e., a chemical or protein modification), wherein the epigenetic variant associated with the presence of a genetic variant such as a cancer-associated mutation.
  • the sample comprises an epigenetic variant associated with the presence of a genetic variant, wherein the sample does not comprise the genetic variant.
  • Exemplary amounts of cell-free nucleic acids in a sample before amplification range from about 1 fg to about 1 pg, e.g., 1 pg to 200 ng, 1 ng to 100 ng, 10 ng to 1000 ng.
  • the amount can be up to about 600 ng, up to about 500 ng, up to about 400 ng, up to about 300 ng, up to about 200 ng, up to about 100 ng, up to about 50 ng, or up to about 20 ng of cell-free nucleic acid molecules.
  • the amount can be at least 1 fg, at least 10 fg, at least 100 fg, at least 1 pg, at least 10 pg, at least 100 pg, at least 1 ng, at least 10 ng, at least 100 ng, at least 150 ng, or at least 200 ng of cell-free nucleic acid molecules.
  • the amount can be up to 1 femtogram (fg), 10 fg, 100 fg, 1 picogram (pg), 10 pg, 100 pg, 1 ng, 10 ng, 100 ng, 150 ng, or 200 ng of cell-free nucleic acid molecules.
  • the method can comprise obtaining 1 femtogram (fg) to 200 ng- [0326]
  • Cell-free nucleic acids are nucleic acids not contained within or otherwise bound to a cell or in other words nucleic acids remaining in a sample after removing intact cells.
  • Cell- free nucleic acids include DNA, RNA, and hybrids thereof, including genomic DNA, mitochondrial DNA, siRNA, miRNA, circulating RNA (cRNA), tRNA, rRNA, small nucleolar RNA (snoRNA), Piwi-interacting RNA (piRNA), long non-coding RNA (long ncRNA), or fragments of any of these.
  • Cell-free nucleic acids can be double-stranded, singlestranded, or a hybrid thereof.
  • a cell-free nucleic acid can be released into bodily fluid through secretion or cell death processes, e.g., cellular necrosis and apoptosis.
  • Some cell-free nucleic acids are released into bodily fluid from cancer cells e.g., circulating tumor DNA, (ctDNA). Others are released from healthy cells.
  • cfDNA is cell-free fetal DNA (cffDNA)
  • cell free nucleic acids are produced by tumor cells.
  • cell free nucleic acids are produced by a mixture of tumor cells and non-tumor cells.
  • Cell-free nucleic acids have an exemplary size distribution of about 100-500 nucleotides, with molecules of 110 to about 230 nucleotides representing about 90% of molecules, with a mode of about 168 nucleotides and a second minor peak in a range between 240 to 440 nucleotides.
  • Molecular tagging refers to a tagging practice that allows one to differentiate among DNA molecules from which sequence reads originated. Tagging strategies can be divided into unique tagging and non-unique tagging strategies. In unique tagging, all or substantially all of the molecules in a sample bear a different tag, so that reads can be assigned to original molecules based on tag information alone. Tags used in such methods are sometimes referred to as “unique tags”. In non-unique tagging, different molecules in the same sample can bear the same tag, so that other information in addition to tag information is used to assign a sequence read to an original molecule. Such information may include start and stop coordinate, coordinate to which the molecule maps, start or stop coordinate alone, etc.
  • the unique tags may be loaded so that less than about 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 50, 100, 500, 1000, 5000, 10000, 50,000, 100,000, 500,000, 1 ,000,000, 10,000,000, 50,000,000 or 1 ,000,000,000 unique tags are loaded per genome sample.
  • the average number of unique tags loaded per sample genome is less than, or greater than, about 1 , 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 50, 100, 500, 1000, 5000, 10000, 50,000, 100,000, 500,000, 1 ,000,000, 10,000,000, 50,000,000 or 1 ,000,000,000 unique tags per genome sample.
  • the capture yield of the target-binding probes specific for the sequence-variable target region set is at least 1.25-, 1.5-, 1.75-, 2-, 2.25- , 2.5-, 2.75-, 3-, 3.5-, 4-, 4.5-, 5-, 6-, 7- 8-, 9-, 10- 11 -, 12-, 13-, 14-, or 15-fold higher than the capture yield of the target-binding probes specific for the epigenetic target region set.
  • concentration may refer to the average mass per volume concentration of individual probes in each set.
  • the target-specific probes specific for the sequencevariable target region set have a higher affinity for their targets than the target-specific probes specific for the epigenetic target region set. Affinity can be modulated in any way known to those skilled in the art, including by using different probe chemistries. For example, certain nucleotide modifications, such as cytosine 5-methylation (in certain sequence contexts), modifications that provide a heteroatom at the T sugar position, and LNA nucleotides, can increase stability of double-stranded nucleic acids, indicating that oligonucleotides with such modifications have relatively higher affinity for their complementary sequences.
  • the target-specific probes are linked to a solid support, e.g., covalently or non-covalently such as through the interaction of a binding pair of capture moieties.
  • the solid support is a bead, such as a magnetic bead.
  • the target-specific probes specific for the sequencevariable target region set and/or the target-specific probes specific for the epigenetic target region set comprise a capture moiety as discussed above, e.g., probes comprising capture moieties and sequences selected to tile across a panel of regions, such as genes.
  • the target-specific probes are provided in a single composition.
  • the single composition may be a solution (liquid or frozen). Alternatively, it may be a lyophilizate.
  • the target-specific probes may be provided as a plurality of compositions, e.g., comprising a first composition comprising probes specific for the epigenetic target region set and a second composition comprising probes specific for the sequence-variable target region set.
  • These probes may be mixed in appropriate proportions to provide a combined probe composition with any of the foregoing fold differences in concentration and/or capture yield.
  • they may be used in separate capture procedures (e.g., with aliquots of a sample or sequentially with the same sample) to provide first and second compositions comprising captured epigenetic target regions and sequence-variable target regions, respectively.
  • Probes specific for epigenetic target regions e.g., comprising a first composition comprising probes specific for the epigenetic target region set and a second composition comprising probes specific for the sequence-variable target region set.
  • the probes for the epigenetic target region set may comprise probes specific for one or more types of target regions likely to differentiate DNA originating from different types of immune cells, including rare immune cell types, and/or to differentiate DNA from precancerous or neoplastic (e.g., tumor or cancer) cells from healthy cells, e.g., non-neoplastic circulating cells. Exemplary types of such regions are discussed in detail herein.
  • the probes for the epigenetic target region set may also comprise probes for one or more control regions, e.g., as described herein.
  • the probes for the epigenetic target region probe set have a footprint of at least 100 kb, e.g., at least 200 kb, at least 300 kb, or at least 400 kb. In some embodiments, the probes for the epigenetic target region set have a footprint in the range of 100- 1000 kb, e.g., 100-200 kb, 200-300 kb, 300-400 kb, 400-500 kb, 500- 600 kb, 600-700 kb, 700- 800 kb, 800-900 kb, and 900-1 ,000 kb. In some embodiments, the probes for the epigenetic target region probe set have a footprint of at least 5 kb, e.g., at least 10, 20, or 50 kb. a. Hypermethylation target regions.
  • the probes can be designed to target either the converted molecules or unconverted molecules depending on the type of methylationsensitive conversion and the target region being enriched. For example, if bisulfite treatment is used, the unmethylated cytosines in the DNA molecules will be converted to dihydrouracil and methylated cytosines will remain unconverted as cytosine.
  • methylation-sensitive conversion e.g., bisulfite or EM-seq
  • the probes can be designed to capture the unconverted molecules, whereas for capturing molecules in the hypomethylated target regions (where the molecules of interest to cancer or any other disease under investigation will be hypomethylated or unmethylated), the probes can be designed to capture the converted molecules
  • the probes for the epigenetic target region set comprise probes specific for one or more hypermethylation target regions.
  • the hypermethylation target regions may be any of those set forth above.
  • the probes specific for hypermethylation target regions comprise probes specific for a plurality of loci that are differentially methylated in different immune cell types.
  • each immune cell type specific hypermethylation target region comprises at least one CpG site that is methylated with a frequency greater than or equal to 0.3, 0.4, 0.5, or 0.6 in one immune cell type and with a frequency less than or equal to 0.1 , 0.2, or 0.3 in all other immune cell types.
  • each immune cell type specific hypermethylation target region comprises at least two CpG sites within 100 base pairs of each other that are each methylated with a frequency greater than or equal to 0.3, 0.4, 0.5, or 0.6 in one immune cell type and with a frequency less than or equal to 0.1 , 0.2, or 0.3 in all other immune cell types.
  • each immune cell type specific hypermethylation target region comprises a total of at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 CpG sites within 150 base pairs or within 200 base pairs, wherein fewer than three of the at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 CpG sites are methylated with a frequency greater than 0.1 , 0.2, or 0.3 in any normal tissue type.
  • each immune cell type specific epigenetic target region set comprises at least 3, at least 5, at least 10, at least 20, or at least 30 hypermethylation target regions that are uniquely hypermethylated in each one of the immune cell types that are identified in the method.
  • the probes specific for hypermethylation target regions comprise probes specific for a plurality of loci listed in Table 1 , e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the loci listed in Table 1.
  • the probes specific for hypermethylation target regions comprise probes specific for a plurality of loci listed in Table 2, e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the loci listed in Table 2.
  • each locus included as a target region there may be one or more probes with a hybridization site that binds between the transcription start site and the stop codon (the last stop codon for genes that are alternatively spliced) of the gene.
  • the one or more probes bind within 300 bp of the listed position, e.g., within 200 or 100 bp.
  • a probe has a hybridization site overlapping the position listed above.
  • the probes specific for the hypermethylation target regions include probes specific for one, two, three, four, or five subsets of hypermethylation target regions that collectively show hypermethylation in one, two, three, four, or five of breast, colon, kidney, liver, and lung cancers, b. Hypomethylation target regions.
  • the probes for the epigenetic target region set comprise probes specific for one or more hypomethylation target regions.
  • the hypomethylation target regions may be any of those set forth above.
  • the probes specific for hypomethylation target regions comprise probes specific for a plurality of loci that are differentially methylated in different immune cell types.
  • each immune cell type specific hypomethylation target region comprises at least one CpG site that is methylated with a frequency less than or equal to 0.1 , 0.2, or 0.3 in one immune cell type and with a frequency greater than or equal to 0.3, 0.4, 0.5, or 0.6 in all other immune cell types.
  • each immune cell type specific hypomethylation target region comprises at least two CpG sites within 100 base pairs of each other that are each methylated with a frequency less than or equal to 0.1 , 0.2, or 0.3 in one immune cell type and with a frequency greater than or equal to 0.3, 0.4, 0.5, or 0.6 in all other immune cell types.
  • each immune cell type specific hypomethylation target region comprises a total of at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 CpG sites within 150 base pairs or within 200 base pairs, wherein fewer than three of the at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 CpG sites are methylated with a frequency less than 0.1 , 0.2, or 0.3 in any normal tissue type.
  • each immune cell type specific epigenetic target region set comprises at least 3, at least 5, at least 10, at least 20, or at least 30 hypomethylation target regions that are uniquely hypomethylated in each one of the immune cell types that are identified in the method.
  • the probes specific for one or more hypomethylation target regions may include probes for regions such as repeated elements, e.g., LINE1 elements, Alu elements, centromeric tandem repeats, pericentromeric tandem repeats, and satellite DNA, and intergenic regions that are ordinarily methylated in healthy cells may show reduced methylation in tumor cells.
  • regions such as repeated elements, e.g., LINE1 elements, Alu elements, centromeric tandem repeats, pericentromeric tandem repeats, and satellite DNA, and intergenic regions that are ordinarily methylated in healthy cells may show reduced methylation in tumor cells.
  • probes specific for hypomethylation target regions include probes specific for repeated elements and/or intergenic regions.
  • probes specific for repeated elements include probes specific for one, two, three, four, or five of LINE1 elements, Alu elements, centromeric tandem repeats, pericentromeric tandem repeats, and/or satellite DNA.
  • the probes specific for focal amplifications include probes specific for one or more of AR, BRAF, CCND1 , CCND2, CCNE1 , CDK4, CDK6, EGFR, ERBB2, FGFR1 , FGFR2, KIT, KRAS, MET, MYC, PDGFRA, PIK3CA, and RAFI.
  • the probes specific for focal amplifications include probes specific for one or more of at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 , 12, 13, 14, 15, 16, 17, or 18 of the foregoing targets.
  • the probes specific for the epigenetic target region set include probes specific for positive control regions that are expected to be methylated in essentially all samples. In some embodiments, the probes specific for the epigenetic target region set include probes specific for negative control regions that are expected to be hypomethylated or unmethylated in essentially all samples.
  • the probes for the sequence-variable target region set may comprise probes specific for a plurality of regions known to undergo somatic mutations in cancer.
  • the probes may be specific for any sequence-variable target region set described herein. Exemplary sequence-variable target region sets are discussed in detail herein, e.g., in the sections above concerning captured sets.
  • the sequence-variable target region probe set has a footprint of at least 0.5 kb, e.g., at least 1 kb, at least 2 kb, at least 5 kb, at least 10 kb, at least 20 kb, at least 30 kb, or at least 40 kb.
  • the epigenetic target region probe set has a footprint in the range of 0.5-100 kb, e.g., 0.5-2 kb, 2-10 kb, 10-20 kb, 20-30 kb, 30-40 kb, 40-50 kb, 50- 60 kb, 60-70 kb, 70-80 kb, 80-90 kb, and 90-100 kb.
  • probes specific for the sequence-variable target region set comprise probes specific for at least a portion of at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, or at 70 of the genes of Table 4.
  • probes specific for the sequence- variable target region set comprise probes specific for the at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, or 70 of the SNVs of Table 3.
  • probes specific for the sequence-variable target region set comprise probes specific for at least 1 , at least 2, at least 3, at least 4, at least 5, or 6 of the fusions of Table 3. In some embodiments, probes specific for the sequence-variable target region set comprise probes specific for at least a portion of at least 1 , at least 2, or 3 of the indels of Table 4. In some embodiments, probes specific for the sequencevariable target region set comprise probes specific for at least a portion of at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, at least 70, or 73 of the genes of Table 5.
  • probes specific for the sequence-variable target region set comprise probes specific for at least a portion of at least 1 , at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11 , at least 12, at least 13, at least 14, at least 15, at least 16, at least 17, or 18 of the indels of Table 5.
  • probes specific for the sequence-variable target region set comprise probes specific for at least a portion of at least 1 , at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11 , at least 12, at least 13, at least 14, at least 15, at least 16, at least 17, at least 18, at least 19, or at least 20 of the genes of Table 6.
  • the probes specific for the sequence-variable target region set comprise probes specific for target regions from at least 10, 20, 30, or 35 cancer-related genes, such as AKTI, ALK, BRAF, CCND1 , CDK2A, CTNNB1 , EGFR, ERBB2, ESR1 , FGFR1 , FGFR2, FGFR3, FOXL2, GAT A3, GNA11 , GNAQ, GNAS, HRAS, IDH1 , IDH2, KIT, KRAS, MED 12, MET, MYC, NFE2L2, NRAS, PDGFRA, PIK3CA, PPP2R1A, PTEN, RET, STK11 , TP53, and U2AF 1.
  • cancer-related genes such as AKTI, ALK, BRAF, CCND1 , CDK2A, CTNNB1 , EGFR, ERBB2, ESR1 , FGFR1 , FGFR2, FGFR3, FOXL2, GAT A3,
  • the precision diagnostics provided by the improved computer system 110 may result in precision treatment plans, which may be identified by the computer system 110 (and/or curated by health professionals).
  • precision treatment plans may relate to genes in the homologous recombination repair (HRR) pathway.
  • HRR homologous recombination repair
  • the number and types of variant nucleotides in a sample can provide an indication of the amenability of the subject providing the sample to treatment, i.e., therapeutic intervention.
  • various poly ADP ribose polymerase (PARP) inhibitors have been shown to stop the growth of tumors from breast, ovarian and prostate cancers caused by hereditary mutations in the BRCA1 or BRCA2 genes.
  • Some of these therapeutic agents may inhibit base excision repair (BER), which may compensate for the deficiency of HRR.
  • a PARP inhibitor may be administered to an individual harboring a somatic homozygous deletion in a HRR gene, but not to an individual harboring a wildtype allele or somatic heterozygous deletions in the HRR gene.
  • a subject having HRD as determined by any of the methods disclosed may be administered a targeted therapy.
  • the targeted therapy may comprise a PARP inhibitor.
  • PARP inhibitors that may be administered include one or more of: VELIPARIB, OLAPARIB, TALAZOPARIB, RUCAPARIB, NIRAPARIB, PAMIPARIB, CEP 9722 (Cephalon), E7016 (Eisai), E7449 (Eisai, a PARP 1/2 and tankyrase 1/2 inhibitor), or 3-Aminobenzamide.
  • the targeted therapy may comprise at least one base excision repair (BER) inhibitor.
  • BER base excision repair
  • OLAPARIB may inhibit BER.
  • the targeted therapy may comprise combination of a PARP inhibitor and radiotherapy.
  • the combination of a PARP inhibitor and radiotherapy would permit the PARP inhibitor to lead to formation of double strand breaks from the single-strand breaks generated by the radiotherapy in tumor tissue (e.g., tissue with BRCA1/BRCA2 mutations). This combination can provide more powerful therapy per radiation dose.
  • tumor tissue e.g., tissue with BRCA1/BRCA2 mutations.
  • the methods disclosed herein relate to identifying and administering therapies, such as customized therapies, to patients or subjects based on the determination of the presence or absence or levels of epigenomic and/or genetic variation.
  • the patient or subject has a given disease, disorder or condition, e.g., any of the cancers or other conditions described elsewhere herein.
  • any cancer therapy e.g., surgical therapy, radiation therapy, chemotherapy, immunotherapy, and/or the like
  • the disease under consideration is a type of cancer.
  • cancers include biliary tract cancer, bladder cancer, transitional cell carcinoma, urothelial carcinoma, brain cancer, gliomas, astrocytomas, breast cancer, metaplastic carcinoma, cervical cancer, cervical squamous cell carcinoma, rectal cancer, colorectal carcinoma, colon cancer, hereditary nonpolyposis colorectal cancer, colorectal adenocarcinomas, gastrointestinal stromal tumors (GISTs), endometrial carcinoma, endometrial stromal sarcomas, esophageal cancer, esophageal squamous cell carcinoma, esophageal adenocarcinoma, ocular melanoma, uveal melanoma, gallbladder carcinomas, gallbladder adenocarcinoma, renal cell carcinoma, clear cell renal cell carcinoma, transitional cell carcinoma, urothelial carcinomas, Wilm
  • Non-limiting examples of other genetic-based diseases, disorders, or conditions that are optionally evaluated using the methods and systems disclosed herein include achondroplasia, alpha- 1 antitrypsin deficiency, antiphospholipid syndrome, autism, autosomal dominant polycystic kidney disease, Charcot-Marie-Tooth (CMT), ch du chat, Crohn's disease, cystic fibrosis, Dercum disease, down syndrome, Duane syndrome, Duchenne muscular dystrophy, Factor V Leiden thrombophilia, familial hypercholesterolemia, familial mediterranean fever, fragile X syndrome, Gaucher disease, hemochromatosis, hemophilia, holoprosencephaly, Huntington's disease, Klinefelter syndrome, Marfan syndrome, myotonic dystrophy, neurofibromatosis, Noonan syndrome, osteogenesis imperfecta, Parkinson's disease, phenylketonuria, Poland anomaly, porphyria, progeria, retin
  • the therapy administered to a subject comprises at least one chemotherapy drug.
  • the chemotherapy drug may comprise alkylating agents (for example, but not limited to, Chlorambucil, Cyclophosphamide, Cisplatin and Carboplatin), nitrosoureas (for example, but not limited to, Carmustine and Lomustine), anti-metabolites (for example, but not limited to, Fluorauracil, Methotrexate and Fludarabine), plant alkaloids and natural products (for example, but not limited to, Vincristine, Paclitaxel and Topotecan), anti- tumor antibiotics (for example, but not limited to, Bleomycin, Doxorubicin and Mitoxantrone), hormonal agents (for example, but not limited to, Prednisone, Dexamethasone, Tamoxifen and Leuprolide) and biological response modifiers (for example, but not limited to, Herceptin and Avastin, Erbitux and Rituxan
  • the chemotherapy administered to a subject may comprise FOLFOX or FOLFIRI.
  • a therapy may be administered to a subject that comprises at least one PARP inhibitor.
  • the PARP inhibitor may include OLAPARIB, TALAZOPARIB, RUCAPARIB, NIRAPARIB (trade name ZEJULA), among others.
  • the methods comprise administering a therapy comprising a PARP inhibitor, such as olaparib, to a subject determined to have homologous recombination repair (HRR) gene or deficiency (HRD), such as with BRCA1 , BRCA2, ATM, BARD1 , BRIP1 , CDK12, CHEK1 , CHEK2, FANCL, PALB2, RAD51 B, RAD51 C, RAD51 D, and RAD54L alterations.
  • the subject has a metastatic castrate resistant prostate cancer (mCRPC).
  • the PARP inhibitor such as olaprib is used to treat a subject having ovarian cancer, breast cancer, pancreatic cancer, or mCRPC, wherein the subject is determined to have alterations in BRCA1 , BRCA2, and/or ATM.
  • Customized therapies can include at least one immunotherapy (or an immunotherapeutic agent).
  • Immunotherapy refers generally to methods of enhancing an immune response against a given cancer type.
  • immunotherapy refers to methods of enhancing a T cell response against a tumor or cancer.
  • the immunotherapy or immunotherapeutic agent targets an immune checkpoint molecule.
  • Certain tumors are able to evade the immune system by co-opting an immune checkpoint pathway.
  • targeting immune checkpoints has emerged as an effective approach for countering a tumor’s ability to evade the immune system and activating anti-tumor immunity against certain cancers. Pardoll, Nature Reviews Cancer, 2012, 12:252-264.
  • the immune checkpoint molecule is an inhibitory molecule that reduces a signal involved in the T cell response to antigen.
  • CTLA4 is expressed on T cells and plays a role in downregulating T cell activation by binding to CD80 (aka B7.1 ) or CD86 (aka B7.2) on antigen presenting cells.
  • PD-1 is another inhibitory checkpoint molecule that is expressed on T cells. PD-1 limits the activity of T cells in peripheral tissues during an inflammatory response.
  • the ligand for PD-1 (PD-L1 or PD-L2) is commonly upregulated on the surface of many different tumors, resulting in the downregulation of anti-tumor immune responses in the tumor microenvironment.
  • the inhibitory immune checkpoint molecule is CTLA4 or PD-1.
  • the inhibitory immune checkpoint molecule is a ligand for PD-1 , such as PD-L1 or PD-L2.
  • the inhibitory immune checkpoint molecule is a ligand for CTLA4, such as CD80 or CD86.
  • the inhibitory immune checkpoint molecule is lymphocyte activation gene 3 (LAG3), killer cell immunoglobulin like receptor (KIR), T cell membrane protein 3 (TIM3), galectin 9 (GAL9), or adenosine A2a receptor (A2aR).
  • the immunotherapy or immunotherapeutic agent is an antagonist of an inhibitory immune checkpoint molecule.
  • the inhibitory immune checkpoint molecule is PD-1.
  • the inhibitory immune checkpoint molecule is PD-L1.
  • the antagonist of the inhibitory immune checkpoint molecule is an antibody (e.g., a monoclonal antibody).
  • the antibody or monoclonal antibody is an anti-CTLA4, anti-PD-1 , anti-PD- L1 , or anti-PD-L2 antibody.
  • the antibody is a monoclonal anti- PD-1 antibody. In some embodiments, the antibody is a monoclonal anti-PD-L1 antibody. In certain embodiments, the monoclonal antibody is a combination of an anti-CTLA4 antibody and an anti-PD-1 antibody, an anti-CTLA4 antibody and an anti-PD-L1 antibody, or an anti-PD-L1 antibody and an anti-PD-1 antibody. In certain embodiments, the anti- PD-1 antibody is one or more of pembrolizumab (Keytruda®) or nivolumab (Opdivo®). In certain embodiments, the anti-CTLA4 antibody is ipilimumab (Yervoy®).
  • the anti-PD-L1 antibody is one or more of atezolizumab (Tecentriq®), avelumab (Bavencio®), or durvalumab (Imfinzi®).
  • immunotherapy such as pembrolizumab, is used to treat a subject determined to have a high microsatellite instability status (MSI-H).
  • MSI-H microsatellite instability status
  • the immunotherapy such as pembrolizumab
  • TMB tumor mutational burden
  • the immunotherapy such as pembrolizumab
  • the immunotherapy is used to treat a subject determined to a have a mismatch repair deficiency (dMMR), such as in genes comprising MLH1 , PMS2, MSH2 and MSH6.
  • dMMR mismatch repair deficiency
  • the immunotherapy or immunotherapeutic agent is an antagonist (e.g., antibody) against CD80, CD86, LAG3, KIR, TIM3, GAL9, or A2aR.
  • the antagonist is a soluble version of the inhibitory immune checkpoint molecule, such as a soluble fusion protein comprising the extracellular domain of the inhibitory immune checkpoint molecule and an Fc domain of an antibody.
  • the soluble fusion protein comprises the extracellular domain of CTLA4, PD-1 , PD-L1 , or PD-L2.
  • the soluble fusion protein comprises the extracellular domain of CD80, CD86, LAG3, KIR, TIM3, GAL9, or A2aR.
  • the soluble fusion protein comprises the extracellular domain of PD-L2 or LAG3.
  • the immune checkpoint molecule is a co-stimulatory molecule selected from CD28, inducible T cell co-stimulator (ICOS), CD137, 0X40, or CD27.
  • the immune checkpoint molecule is a ligand of a co-stimulatory molecule, including, for example, CD80, CD86, B7RP1 , B7-H3, B7-H4, CD137L, OX40L, or CD70.
  • Agonists that target these co-stimulatory checkpoint molecules can be used to enhance antigen-specific T cell responses against certain cancers.
  • the immunotherapy or immunotherapeutic agent is an agonist of a co-stimulatory checkpoint molecule.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Informatics (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Theoretical Computer Science (AREA)
  • Biophysics (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Biotechnology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Software Systems (AREA)
  • Public Health (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Epidemiology (AREA)
  • Databases & Information Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioethics (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Proteomics, Peptides & Aminoacids (AREA)
  • Image Analysis (AREA)

Abstract

Images are produced having pixel values based on molecular characteristics and genomic position data corresponding to nucleic acid molecules derived from a sample. The images can be provided as input to a convolutional neural network that analyzes the images to determine a tumor indication related to a tumor being present in one or more subjects.

Description

METHODS FOR CANCER DETECTION USING MOLECULAR PATTERNS
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application claims the benefit of priority of US Provisional Patent Application No. 63/571 ,333, filed March 28, 2024, and US Provisional Patent Application No. 63/651 ,652, filed May 24, 2024, each of which is incorporated by reference herein in its entirety for all purposes
BACKGROUND
[0002] Cancer is a major cause of disease worldwide. Each year, tens of millions of people are diagnosed with cancer around the world, and more than half eventually die from it. In many countries, cancer ranks the second most common cause of death following cardiovascular diseases. Early detection is associated with improved outcomes for many cancers.
[0003] Cancer can be caused by the accumulation of genetic variations within an individual's normal cells, at least some of which result in improperly regulated cell division. Such variations commonly include copy number variations (CNVs), single nucleotide variations (SNVs), gene fusions, insertions and/or deletions (indels), epigenetic variations including 5-methylation of cytosine (5-methylcytosine), and association of DNA with chromatin and transcription factors.
[0004] Cancers are often detected by biopsies of tumors followed by analysis of cell markers or DNA extracted from cells. But more recently it has been proposed that cancers can also be detected from cell-free nucleic acids in body fluids, such as blood or urine. Such tests have the advantage that they are noninvasive and can be performed without identifying suspected cancer cells in biopsy. However, such tests are complicated by the fact that the amount of nucleic acids in body fluids is very low and that the nucleic acids that are present are heterogeneous in form (e.g., RNA and DNA, single-stranded and double-stranded, and various states of post-replication modification and association with proteins, such as histones).
[0005] Thus, there is a need for improved systems and methods for improved cancer detection using liquid biopsy assays. Therefore, it is an object of the disclosure to provide computer-implemented systems and methods and other processes that have improved capability to classify a sample as containing tumor-derived DNA.
BRIEF DESCRIPTION OF THE DRAWINGS
[0006] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate certain implementations, and together with the written description, serve to explain certain principles of the methods, computer readable media, and systems disclosed herein. The description provided herein is better understood when read in conjunction with the accompanying drawings which are included by way of example and not by way of limitation. It will be understood that like reference numerals identify like components throughout the drawings, unless the context indicates otherwise. It will also be understood that some or all of the figures may be schematic representations for purposes of illustration and do not necessarily depict the actual relative sizes or locations of the elements shown.
[0007] Figure 1 is a diagrammatic representation of an example computational architecture that implements one or more convolutional neural networks to identify samples obtained from subjects in which a tumor-related biological condition is present, according to one or more example implementations.
[0008] Figure 2 is a diagrammatic representation of an example computational architecture to generate image data derived from sample data and implementing a number of convolutional neural networks to analyze the image data for the detection of tumor-related biological conditions, according to one or more example implementations. [0009] Figure 3 is a diagrammatic representation of an example computational architecture to implement one or more convolutional neural networks to detect a plurality of cancer types, according to one or more example implementations.
[0010] Figure 4 is a flow diagram of an example process to generate image data and implement one or more convolutional neural networks to analyze the image data for the detection of a tumor-derived biological condition, according to one or more implementations.
[0011] Figure 5 is a block diagram illustrating components of a machine, in the form of a computer system, that may read and execute instructions from one or more machine- readable media to perform any one or more methodologies described herein, in accordance with one or more example implementations.
[0012] Figure 6 is block diagram illustrating a representative software architecture that may be used in conjunction with one or more hardware architectures described herein, in accordance with one or more example implementations.
SUMMARY
[0013] In one or more aspects, the techniques described herein relate to a method including obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a plurality of groups of sequence representations that correspond to a plurality of genomic regions, individual groups of sequence representations of the plurality of groups of sequence representations being aligned with a discrete genomic region; computationally analyzing the individual groups of sequence representations to determine methylation rates of cytosine-guanine dinucleotides included in the individual groups of sequence representations; determining, based on the methylation rates, subsets of sequence representations from among the plurality of groups of sequence representations, individual subsets satisfying one or more methylation rate criteria; generating, based on individual subsets of sequence representations, one or more images for individual genomic regions of the plurality of genomic regions, individual images of the one or more images including a plurality of pixels, wherein individual pixels of the plurality of pixels comprise (i) a first value that corresponds to a genomic location within the individual genomic region, (ii) a second value that corresponds to the methylation rate of sequence representations included in a subset of sequence representations corresponding to the individual genomic region, and (iii) an intensity value indicating a number of the sequence representations included in the subset of sequence representations having the first value and the second value; generating a plurality of convolutional neural networks, individual convolutional networks of the plurality of convolutional networks corresponding to an individual genomic region, wherein the individual convolutional neural networks for the individual genomic regions determine a tumor indication related to a tumor being present in the one or more subjects based on the one or more images that correspond to the individual genomic regions; and computationally analyzing the tumor indications determined by the plurality of convolutional neural networks to determine an overall tumor indication.
[0014] In one or more aspects, the techniques described herein relate to a method including: obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected; generating first training images based on the first training sequence representations, individual first training images including a first plurality of pixels, wherein individual pixels of the first plurality of pixels comprise (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to methylation rate of at least a portion of the first training sequences representations, and (iii) a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value; obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected; generating second training images based on the second training sequence representations, individual second training images including a second plurality of pixels, wherein individual pixels of the second plurality of pixels comprise (i) a first additional training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second additional training value that corresponds to methylation rate of at least a portion of the second training sequences representation, and (iii) a second intensity training value indicating a number of the second training sequence representations having the first additional training value and the second additional training value; and performing a plurality of iterations of a training process for the convolutional neural network to determine weights of layers of the convolutional neural network.
[0015] In one or more aspects, the techniques described herein relate to a method wherein individual iterations of the plurality of iterations include: determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network; determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network; determining differences between the first weights and the second weights; and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights.
[0016] In one or more aspects, the techniques described herein relate to a method wherein the methylation rates of cytosine-guanine dinucleotides included in the individual sequence representations are determined by procedures that affect a first nucleobase differently from a second nucleobase.
[0017] In one or more aspects, the techniques described herein relate to a method including dividing a sample of the one or more samples into a plurality of subsamples including a first subsample corresponding to a first partition and a second subsample corresponding to a second partition, wherein the first partition comprises nucleic acids with a cytosine modification in a greater proportion than additional nucleic acids included in the second partition; wherein the one or more methylation criteria correspond to (first partition?) characteristics correspond the nucleic acids included in the first partition or the additional nucleic acids included in the second partition.
[0018] In one or more aspects, the techniques described herein relate to a method wherein a logistic regression technique is implemented to determine the overall tumor indication based on the tumor indications determined by the plurality of convolutional neural networks.
[0019] In one or more aspects, the techniques described herein relate to a method wherein individual tumor indications determined by the individual convolutional neural networks include probabilities of a tumor being present in the one or more subjects.
[0020] In one or more aspects, the techniques described herein relate to a method wherein the probabilities of a tumor being present in the one or more subjects are analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
[0021] In one or more aspects, the techniques described herein relate to a method wherein the convolutional neural network includes a plurality of output layers with each output layer of the plurality of output layers corresponding to a cancer type of a plurality of cancer types; the convolutional neural network determines a plurality of probabilities of the plurality of cancer types being present in one or more subjects; and for individual output layers, the convolutional neural network generates an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in one or more subjects.
[0022] In one or more aspects, the techniques described herein relate to a method including computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
[0023] In one or more aspects, the techniques described herein relate to a method wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels.
[0024] In one or more aspects, the techniques described herein relate to a method including determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
[0025] In one or more aspects, the techniques described herein relate to a system including one or more hardware processors; and memory storing computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a plurality of groups of sequence representations that correspond to a plurality of genomic regions, individual groups of sequence representations of the plurality of groups of sequence representations being aligned with a discrete genomic region; computationally analyzing the individual groups of sequence representations to determine methylation rates of cytosine-guanine dinucleotides included in the individual groups of sequence representations; determining, based on the methylation rates, subsets of sequence representations from among the plurality of groups of sequence representations, individual subsets satisfying one or more methylation rate criteria; generating, based on individual subsets of sequence representations, one or more images for individual genomic regions of the plurality of genomic regions, individual images of the one or more images including a plurality of pixels, wherein individual pixels of the plurality of pixels comprise (i) a first value that corresponds to a genomic location within the individual genomic region, (ii) a second value that corresponds to the methylation rate of sequence representations included in a subset of sequence representations corresponding to the individual genomic region, and (iii) an intensity value indicating a number of the sequence representations included in the subset of sequence representations having the first value and the second value; generating a plurality of convolutional neural networks, individual convolutional networks of the plurality of convolutional networks corresponding to an individual genomic region, wherein the individual convolutional neural networks for the individual genomic regions determine a tumor indication related to a tumor being present in the one or more subjects based on the one or more images that correspond to the individual genomic regions; and computationally analyzing the tumor indications determined by the plurality of convolutional neural networks to determine an overall tumor indication.
[0026] In one or more aspects, the techniques described herein relate to a system wherein the memory stores additional computer-readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the individual subsets of sequence representations are computational analyzed to generate one or more additional images for the individual genomic regions, wherein the one or more additional images for the individual genomic regions includes first pixel values that comprise (i) first values that correspond to genomic locations within the individual genomic region and (ii) second values that correspond to an additional molecular characteristic of the individual sequence representations of the group of sequence representations. [0027] In one or more aspects, the techniques described herein relate to a system wherein the one or more additional molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the subset of sequence representations corresponding to the individual genomic region, a length of the individual sequence representations of the subset of sequence representations corresponding to the individual genomic region, or a number of restriction enzyme cut sites in the individual sequence representations of the subset of sequence representations corresponding to the individual genomic region.
[0028] In one or more aspects, the techniques described herein relate to a system wherein the one or more methylation rate criteria correspond to at least a threshold number of methylated cytosine-guanine dinucleotides.
[0029] In one or more aspects, the techniques described herein relate to a system wherein the one or more methylation rate criteria correspond to no greater than a threshold number of methylated cytosine-guanine dinucleotides.
[0030] In one or more aspects, the techniques described herein relate to a system wherein the one or more samples are partitioned into a plurality of subsamples on the basis of methylate rate and the one or more methylation rate criteria correspond to a partition of a plurality of partitions into which the plurality of subsamples are divided.
[0031] In one or more aspects, the techniques described herein relate to a method including: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations, the group of sequence representations corresponding to one or more portions of a genomic region; computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations; generating, based on the group of sequence representations, one or more images that include a plurality of pixels, wherein individual pixels of the plurality of pixels include (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value; and providing the one or more images to a convolutional neural network, wherein the convolutional neural network computationally analyzes the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
[0032] In one or more aspects, the techniques described herein relate to a method, including: computationally analyzing the sequencing data to determine a plurality of additional groups of additional sequence representations in relation to a plurality of additional genomic regions; computationally analyzing the plurality of additional groups of additional sequence representations to determine additional values of the one or more molecular characteristics for additional individual sequence representations of the plurality of additional groups of sequence representations; and generating a plurality of additional images based on the plurality of additional groups of sequence representations, wherein: the plurality of additional images include a plurality of additional pixels and individual additional pixels of the plurality of additional pixels include (i) an additional first value that corresponds to one or more additional genomic locations, (ii) an additional second value that corresponds to the one or more molecular characteristics, and (iii) an additional intensity value indicating an additional number of the additional sequence representations having the additional first value and the additional second value.
[0033] In one or more aspects, the techniques described herein relate to a method, wherein: each additional image of the plurality of additional images is generated using information derived from additional sequence representations included in an additional group of sequence representations of the plurality of additional groups of sequence representations and the additional sequence representations are homologous with an additional genomic region.
[0034] In one or more aspects, the techniques described herein relate to a method, including: providing the plurality of additional images to a plurality of additional convolutional neural networks to determine a plurality of additional tumor indications related to a tumor being present in the one or more samples, wherein individual additional convolutional networks of the plurality of additional convolutional neural networks computationally analyze a portion of the plurality of additional images corresponding to a given genomic region to determine an additional tumor indication related to a tumor being present in the one or more subjects; and computationally analyzing the tumor indication and the plurality of additional tumor indications to determine an overall tumor indication related to a tumor being present in the one or more subjects.
[0035] In one or more aspects, the techniques described herein relate to a method, wherein the tumor indication and the plurality of additional tumor indications are computationally analyzed using a logistic regression technique to determine the overall tumor indication.
[0036] In one or more aspects, the techniques described herein relate to a method, wherein the tumor indication and the plurality of additional tumor indications include probabilities of a tumor being present in the one or more subjects.
[0037] In one or more aspects, the techniques described herein relate to a method, wherein the probabilities of a tumor being present in the one or more subjects are computationally analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
[0038] In one or more aspects, the techniques described herein relate to a method, wherein: the convolutional neural network includes a plurality of output layers with each output layer of the plurality of output layers corresponding to a cancer type of a plurality of cancer types; the convolutional neural network determines a plurality of probabilities of the plurality of cancer types being present in one or more subjects; and for individual output layers, the convolutional neural network generates an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in one or more subjects.
[0039] In one or more aspects, the techniques described herein relate to a method, including: computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
[0040] In one or more aspects, the techniques described herein relate to a method, wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels.
[0041] In one or more aspects, the techniques described herein relate to a method, including: determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
[0042] In one or more aspects, the techniques described herein relate to a method, wherein: the one or more images include a first image that corresponds to the genomic region and a second image that corresponds to the genomic region; the first image includes first pixel values that include (i) first values that correspond to genomic locations within the genomic region and (ii) second values that correspond to a first molecular characteristic of the individual sequence representations of the group of sequence representations; and the second image includes second pixel values that include (i) the first values that correspond to the genomic locations within the genomic region and (ii) additional second values that correspond to a second molecular characteristic of the individual sequence representations of the group of sequence representations.
[0043] In one or more aspects, the techniques described herein relate to a method, including: computationally analyzing the first image using a first convolutional neural network to determine a first tumor indication related to a tumor being present in one or more subjects; computationally analyzing the second image using a second convolutional neural network to determine a second tumor indication of a tumor being present in one or more subjects; and determining an overall tumor indication of a tumor being present in one or more subjects based on the first tumor indication and the second tumor indication. [0044] In one or more aspects, the techniques described herein relate to a method, wherein the one or more molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the group of sequence representations, a length of the individual sequence representations of the group of sequence representations, or a number of restriction enzyme cut sites in the individual sequence representations of the group of sequence representations. [0045] In one or more aspects, the techniques described herein relate to a method, including: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
[0046] In one or more aspects, the techniques described herein relate to a method, including: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having no greater than a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
[0047] In one or more aspects, the techniques described herein relate to a method, wherein genomic locations that correspond to first values of the plurality of pixels correspond to an interval that includes a plurality of nucleotides.
[0048] In one or more aspects, the techniques described herein relate to a method, wherein the genomic region is included in a number of genomic regions that are enriched as part of a diagnostic test to determine the presence of tumors in subjects.
[0049] In one or more aspects, the techniques described herein relate to a method, including: obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected; generating first training images based on the first training sequence representations, individual first training images including a first plurality of pixels, wherein individual pixels of the first plurality of pixels include (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to the one or more molecular characteristics, and (iii) a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value; obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected; generating second training images based on the second training sequence representations, individual second training images including a second plurality of pixels, wherein individual pixels of the second plurality of pixels include (i) a first additional training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second additional training value that corresponds to the one or more molecular characteristics, and (iii) a second intensity training value indicating a number of the second training sequence representations having the first additional training value and the second additional training value; and performing a plurality of iterations of a training process for the convolutional neural network to determine weights of layers of the convolutional neural network.
[0050] In one or more aspects, the techniques described herein relate to a method, wherein individual iterations of the plurality of iterations include: determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network; determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network; determining differences between the first weights and the second weights; and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights.
[0051] In one or more aspects, the techniques described herein relate to a method including: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations, the group of sequence representations corresponding to one or more portions of a genomic region; computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations; generating, based on the group of sequence representations, one or more images that include a plurality of pixels, wherein individual pixels of the plurality of pixels include (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value; and providing the one or more images to a convolutional neural network, wherein the convolutional neural network computationally analyzes the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
[0052] In one or more aspects, the techniques described herein relate to one or more computing apparatuses, including: one or more hardware processors; and memory storing computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations, the group of sequence representations corresponding to one or more portions of a genomic region; computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations; generating, based on the group of sequence representations, one or more images that include a plurality of pixels, wherein individual pixels of the plurality of pixels include (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value; and providing the one or more images to a convolutional neural network, wherein the convolutional neural network computationally analyzes the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
[0053] In one or more aspects, the techniques described herein relate to one or more computing apparatuses, including: one or more hardware processors; and memory storing computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising: computationally analyzing the sequencing data to determine a plurality of additional groups of additional sequence representations in relation to a plurality of additional genomic regions; computationally analyzing the plurality of additional groups of additional sequence representations to determine additional values of the one or more molecular characteristics for additional individual sequence representations of the plurality of additional groups of sequence representations; and generating a plurality of additional images based on the plurality of additional groups of sequence representations, wherein: the plurality of additional images include a plurality of additional pixels and individual additional pixels of the plurality of additional pixels include (i) an additional first value that corresponds to one or more additional genomic locations, (ii) an additional second value that corresponds to the one or more molecular characteristics, and (iii) an additional intensity value indicating an additional number of the additional sequence representations having the additional first value and the additional second value.
[0054] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein: each additional image of the plurality of additional images is generated using information derived from additional sequence representations included in an additional group of sequence representations of the plurality of additional groups of sequence representations and the additional sequence representations are homologous with an additional genomic region.
[0055] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the memory stores additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: providing the plurality of additional images to a plurality of additional convolutional neural networks to determine a plurality of additional tumor indications related to a tumor being present in the one or more samples, wherein individual additional convolutional networks of the plurality of additional convolutional neural networks computationally analyze a portion of the plurality of additional images corresponding to a given genomic region to determine an additional tumor indication related to a tumor being present in the one or more subjects; and computationally analyzing the tumor indication and the plurality of additional tumor indications to determine an overall tumor indication related to a tumor being present in the one or more subjects. [0056] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the tumor indication and the plurality of additional tumor indications are computationally analyzed using a logistic regression technique to determine the overall tumor indication.
[0057] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the tumor indication and the plurality of additional tumor indications include probabilities of a tumor being present in the one or more subjects.
[0058] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the probabilities of a tumor being present in the one or more subjects are computationally analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
[0059] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein: the convolutional neural network includes a plurality of output layers with each output layer of the plurality of output layers corresponding to a cancer type of a plurality of cancer types; the convolutional neural network determines a plurality of probabilities of the plurality of cancer types being present in one or more subjects; and for individual output layers, the convolutional neural network generates an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in one or more subjects.
[0060] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the memory stores additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
[0061] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels. [0062] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the memory stores additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
[0063] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein: the one or more images include a first image that corresponds to the genomic region and a second image that corresponds to the genomic region; the first image includes first pixel values that include (i) first values that correspond to genomic locations within the genomic region and (ii) second values that correspond to a first molecular characteristic of the individual sequence representations of the group of sequence representations; and the second image includes second pixel values that include (i) the first values that correspond to the genomic locations within the genomic region and (ii) additional second values that correspond to a second molecular characteristic of the individual sequence representations of the group of sequence representations.
[0064] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the memory stores additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the first image using a first convolutional neural network to determine a first tumor indication related to a tumor being present in one or more subjects; computationally analyzing the second image using a second convolutional neural network to determine a second tumor indication of a tumor being present in one or more subjects; and determining an overall tumor indication of a tumor being present in one or more subjects based on the first tumor indication and the second tumor indication. [0065] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the one or more molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the group of sequence representations, a length of the individual sequence representations of the group of sequence representations, or a number of restriction enzyme cut sites in the individual sequence representations of the group of sequence representations.
[0066] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the memory stores additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
[0067] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the memory stores additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having no greater than a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
[0068] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein genomic locations that correspond to first values of the plurality of pixels correspond to an interval that includes a plurality of nucleotides.
[0069] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the genomic region is included in a number of genomic regions that are enriched as part of a diagnostic test to determine the presence of tumors in subjects.
[0070] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein the memory stores additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected; generating first training images based on the first training sequence representations, individual first training images including a first plurality of pixels, wherein individual pixels of the first plurality of pixels include (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to the one or more molecular characteristics, and (iii) a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value; obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected; generating second training images based on the second training sequence representations, individual second training images including a second plurality of pixels, wherein individual pixels of the second plurality of pixels include (i) a first additional training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second additional training value that corresponds to the one or more molecular characteristics, and (iii) a second intensity training value indicating a number of the second training sequence representations having the first additional training value and the second additional training value; and performing a plurality of iterations of a training process for the convolutional neural network to determine weights of layers of the convolutional neural network.
[0071] In one or more aspects, the techniques described herein relate to one or more computer apparatuses, wherein individual iterations of the plurality of iterations include: determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network; determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network; determining differences between the first weights and the second weights; and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights.
[0072] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media storing computer-readable instructions that, when executed by one or more hardware processors, cause the one or more hardware processors to perform operations comprising : obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations, the group of sequence representations corresponding to one or more portions of a genomic region; computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations; generating, based on the group of sequence representations, one or more images that include a plurality of pixels, wherein individual pixels of the plurality of pixels include (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value; and providing the one or more images to a convolutional neural network, wherein the convolutional neural network computationally analyzes the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
[0073] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media storing computer-readable instructions that, when executed by one or more hardware processors, cause the one or more hardware processors to perform operations comprising: computationally analyzing the sequencing data to determine a plurality of additional groups of additional sequence representations in relation to a plurality of additional genomic regions; computationally analyzing the plurality of additional groups of additional sequence representations to determine additional values of the one or more molecular characteristics for additional individual sequence representations of the plurality of additional groups of sequence representations; and generating a plurality of additional images based on the plurality of additional groups of sequence representations, wherein: the plurality of additional images include a plurality of additional pixels and individual additional pixels of the plurality of additional pixels include (i) an additional first value that corresponds to one or more additional genomic locations, (ii) an additional second value that corresponds to the one or more molecular characteristics, and (iii) an additional intensity value indicating an additional number of the additional sequence representations having the additional first value and the additional second value.
[0074] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein: each additional image of the plurality of additional images is generated using information derived from additional sequence representations included in an additional group of sequence representations of the plurality of additional groups of sequence representations and the additional sequence representations are homologous with an additional genomic region.
[0075] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: providing the plurality of additional images to a plurality of additional convolutional neural networks to determine a plurality of additional tumor indications related to a tumor being present in the one or more samples, wherein individual additional convolutional networks of the plurality of additional convolutional neural networks computationally analyze a portion of the plurality of additional images corresponding to a given genomic region to determine an additional tumor indication related to a tumor being present in the one or more subjects; and computationally analyzing the tumor indication and the plurality of additional tumor indications to determine an overall tumor indication related to a tumor being present in the one or more subjects. [0076] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein the tumor indication and the plurality of additional tumor indications are computationally analyzed using a logistic regression technique to determine the overall tumor indication.
[0077] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein the tumor indication and the plurality of additional tumor indications include probabilities of a tumor being present in the one or more subjects.
[0078] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein the probabilities of a tumor being present in the one or more subjects are computationally analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
[0079] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein: the convolutional neural network includes a plurality of output layers with each output layer of the plurality of output layers corresponding to a cancer type of a plurality of cancer types; the convolutional neural network determines a plurality of probabilities of the plurality of cancer types being present in one or more subjects; and for individual output layers, the convolutional neural network generates an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in one or more subjects.
[0080] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
[0081] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels.
[0082] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
[0083] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein: the one or more images include a first image that corresponds to the genomic region and a second image that corresponds to the genomic region; the first image includes first pixel values that include (i) first values that correspond to genomic locations within the genomic region and (ii) second values that correspond to a first molecular characteristic of the individual sequence representations of the group of sequence representations; and the second image includes second pixel values that include (i) the first values that correspond to the genomic locations within the genomic region and (ii) additional second values that correspond to a second molecular characteristic of the individual sequence representations of the group of sequence representations.
[0084] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the first image using a first convolutional neural network to determine a first tumor indication related to a tumor being present in one or more subjects; computationally analyzing the second image using a second convolutional neural network to determine a second tumor indication of a tumor being present in one or more subjects; and determining an overall tumor indication of a tumor being present in one or more subjects based on the first tumor indication and the second tumor indication.
[0085] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein the one or more molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the group of sequence representations, a length of the individual sequence representations of the group of sequence representations, or a number of restriction enzyme cut sites in the individual sequence representations of the group of sequence representations.
[0086] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
[0087] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having no greater than a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
[0088] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein genomic locations that correspond to first values of the plurality of pixels correspond to an interval that includes a plurality of nucleotides. [0089] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein the genomic region is included in a number of genomic regions that are enriched as part of a diagnostic test to determine the presence of tumors in subjects.
[0090] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media storing additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected; generating first training images based on the first training sequence representations, individual first training images including a first plurality of pixels, wherein individual pixels of the first plurality of pixels include (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to the one or more molecular characteristics, and (iii) a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value; obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected; generating second training images based on the second training sequence representations, individual second training images including a second plurality of pixels, wherein individual pixels of the second plurality of pixels include (i) a first additional training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second additional training value that corresponds to the one or more molecular characteristics, and (iii) a second intensity training value indicating a number of the second training sequence representations having the first additional training value and the second additional training value; and performing a plurality of iterations of a training process for the convolutional neural network to determine weights of layers of the convolutional neural network.
[0091] In one or more aspects, the techniques described herein relate to one or more non-transitory computer-readable media, wherein individual iterations of the plurality of iterations include: determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network; determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network; determining differences between the first weights and the second weights; and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights.
[0092] Other technical features may be readily apparent to one skilled in the art from the following figures, descriptions, and claims.
DEFINITIONS
[0093] In order for the present disclosure to be more readily understood, certain terms are first defined below. Additional definitions for the following terms and other terms may be set forth through the specification. If a definition of a term set forth below is inconsistent with a definition in an application or patent that is incorporated by reference, the definition set forth in this application should be used to understand the meaning of the term.
[0094] As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” include plural references unless the context clearly dictates otherwise. Thus, for example, a reference to “a method” includes one or more methods, and/or steps of the type described herein and/or which will become apparent to those persons of ordinary skill in the art upon reading this disclosure and so forth.
[0095] It is also to be understood that the terminology used herein is for the purpose of describing particular implementations only and is not intended to be limiting. Further, unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. In describing and claiming the methods, computer readable media, and systems, the following terminology, and grammatical variants thereof, will be used in accordance with the definitions set forth below.
[0096]About As used herein, “about” or “approximately” as applied to one or more values or elements of interest, refers to a value or element that is similar to a stated reference value or element. In certain implementations, the term “about” or “approximately” refers to a range of values or elements that falls within 25%, 20%, 19%, 18%, 17%, 16%, 15%, 14%, 13%, 12%, 11 %, 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1 %, or less in either direction (greater than or less than) of the stated reference value or element unless otherwise stated or otherwise evident from the context (except where such number would exceed 100% of a possible value or element).
[0097] Administer: As used herein, “administer” or “administering” a therapeutic agent (e.g., an immunological therapeutic agent) to a subject means to give, apply or bring the composition into contact with the subject. Administration can be accomplished by any of a number of routes, including, for example, topical, oral, subcutaneous, intramuscular, intraperitoneal, intravenous, intrathecal and intradermal.
[0098] Adapter As used herein, “adapter” refers to a short nucleic acid (e.g., less than about 500 nucleotides, less than about 100 nucleotides, or less than about 50 nucleotides in length) that can be at least partially double-stranded and used to link to either or both ends of a given sample nucleic acid molecule. Adapters can include sequences of nucleic acid primer binding sites to permit amplification of a nucleic acid molecule flanked by adapters at both ends, and/or a sequencing primer binding site, including primer binding sites for sequencing applications, such as various next-generation sequencing (NGS) applications. Adapters can also include binding sites for capture probes, such as an oligonucleotide attached to a flow cell support or the like. Adapters can also include a nucleic acid tag as described herein. Nucleic acid tags can be positioned relative to amplification primer and sequencing primer binding sites, such that a nucleic acid tag is included in amplicons and sequence reads of a given nucleic acid molecule. The same or different adapters can be linked to the respective ends of a nucleic acid molecule. In some implementations, the same adapter is linked to the respective ends of the nucleic acid molecule except that the nucleic acid tag differs. In some implementations, the adapter is a Y-shaped adapter in which one end is blunt ended or tailed as described herein, for joining to a nucleic acid molecule, which is also blunt ended or tailed with one or more complementary nucleotides. In still other example implementations, an adapter is a bell-shaped adapter that includes a blunt or tailed end for joining to a nucleic acid molecule to be analyzed. Other examples of adapters include T-tailed and C-tailed adapters.
[0099]Alignment As used herein, “alignment” or “align” refers to determining whether at least two sequence representations have at least a threshold amount of homology. In one or more examples, the threshold amount of homology can be at least about 90%, at least about 91 %, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, at least about 99.5%, or at least about 99.9%. In situations where two sequence representations have at least the threshold amount of homology, the two sequence representations can be referred to as being “aligned.” In various examples, alignment can include determining whether a sequence representation has at least a threshold amount of homology with respect to a reference sequence.
[00100] Amplify. As used herein, “amplify” or “amplification” in the context of nucleic acids refers to the production of multiple copies of a polynucleotide, or a portion of the polynucleotide, starting from a small amount of the polynucleotide (e.g., a single polynucleotide molecule), where the amplification products or amplicons are generally detectable. Amplification of polynucleotides encompasses a variety of chemical and enzymatic processes.
[00101] Barcode. As used herein, “barcode” or “molecular barcode” in the context of nucleic acids refers to a nucleic acid molecule comprising a sequence that can serve as a molecular identifier. For example, individual "barcode" sequences can be added to each DNA fragment during next -generation sequencing (NGS) library preparation so that each read can be identified and sorted before the final data analysis. In one or more examples, individual barcode sequences can be added to DNA fragments during NGS library preparation so that reads corresponding to each unique molecule included in a sample can be identified.
[00102] Cancer Type. As used herein, “cancer type” refers to a type or subtype of cancer defined, e.g., by histopathology. Cancer type can be defined by any conventional criterion, such as on the basis of occurrence in a given tissue (e.g., blood cancers, central nervous system (CNS), brain cancers, lung cancers (small cell and non-small cell), skin cancers, nose cancers, throat cancers, liver cancers, bone cancers, lymphomas, pancreatic cancers, bowel cancers, rectal cancers, thyroid cancers, bladder cancers, kidney cancers, mouth cancers, stomach cancers, breast cancers, prostate cancers, ovarian cancers, lung cancers, intestinal cancers, soft tissue cancers, neuroendocrine cancers, gastroesophageal cancers, head and neck cancers, gynecological cancers, colorectal cancers, urothelial cancers, solid state cancers, heterogeneous cancers, homogenous cancers), unknown primary origin and the like, and/or of the same cell lineage (e.g., carcinoma, sarcoma, lymphoma, cholangiocarcinoma, leukemia, mesothelioma, melanoma, or glioblastoma) and/or cancers exhibiting cancer markers, such as Her2, CA15-3, CA19-9, CA-125, CEA, AFP, PSA, HCG, EGFR, KRAS, APC, RB1 , hormone receptor, estrogen receptor (ER), progesterone receptor (PR), and NMP- 22. Cancers can also be classified by stage (e.g., stage 1 , 2, 3, or 4), cell morphology (e.g., small vs. non-small cell lung cancer), and whether of primary or secondary origin.
[00103] Carrier Signal. As used herein, “carrier signal” refers to any intangible medium that is capable of storing, encoding, or carrying transitory or non-transitory instructions for execution by a machine, and includes digital or analog communications signals or other intangible medium to facilitate communication of such instructions. Instructions may be transmitted or received over a network using a transitory or non- transitory transmission medium via a network interface device and using any one of a number of data transfer protocols.
[00104] Cell-Free Nucleic Acid As used herein, “cell-free nucleic acid” refers to nucleic acids not contained within or otherwise bound to a cell or, in some implementations, nucleic acids remaining in a sample following the removal of intact cells. Cell-free nucleic acids can include, for example, all non-encapsulated nucleic acids sourced from a bodily fluid (e.g., blood, plasma, serum, urine, cerebrospinal fluid (CSF), etc.) from a subject. Cell-free nucleic acids include DNA (cfDNA), RNA (cfRNA), and hybrids thereof, including genomic DNA, mitochondrial DNA, circulating DNA, siRNA, miRNA, circulating RNA (cRNA), tRNA, rRNA, small nucleolar RNA (snoRNA), Piwi- interacting RNA (piRNA), long non-coding RNA (long ncRNA), and/or fragments of any of these. Cell-free nucleic acids can be double-stranded, single-stranded, or a hybrid thereof. A cell-free nucleic acid can be released into bodily fluid through secretion or cell death processes, e.g., cellular necrosis, apoptosis, or the like. Some cell-free nucleic acids are released into bodily fluid from cancer cells, e.g., circulating tumor DNA (ctDNA). Others are released from healthy cells. CtDNA can be non-encapsulated tumor-derived fragmented DNA. A cell-free nucleic acid can have one or more epigenetic modifications, for example, a cell-free nucleic acid can be acetylated, 5-methylated, ubiquitylated, phosphorylated, sumoylated, ribosylated, and/or citrull inated.
[00105] Cellular Nucleic Acids As used herein, “cellular nucleic acids” means nucleic acids that are disposed within one or more cells at least at the point a sample is taken or collected from a subject, even if those nucleic acids are subsequently removed as part of a given analytical process.
[00106] Classification Region. As used herein, “classification region” refers to a genomic region that may show sequence-independent changes in neoplastic cells (e.g., tumor cells and cancer cells) or that may show sequence-independent changes in cfDNA from subjects having cancer relative to cfDNA from subjects in which cancer is not present. Examples of sequence-independent changes include, but are not limited to, changes in methylation rate (increases or decreases), nucleosome distribution, CTCF binding, transcription start sites, and regulatory protein binding regions. In one or more examples, sequence-independent changes in a classification region can indicate the presence of a single form of cancer in a subject. In one or more additional examples, sequence-independent changes in a classification region can correspond to the presence of multiple forms in a subject. The classification region can be enriched by one or more probes. In addition, the classification region can be defined by a pair of primer binding sites. Further, the classification region can be defined by a predetermined beginning genomic locus and a predetermined ending genomic locus. In one or more examples, the genomic locus can correspond to at least one of one or more genomic positions or one or more genomic coordinates. In various examples, the classification region can include nucleotide sequences corresponding to one or more genomic regions that are to be amplified during one or more nucleic acid sequencing processes. The classification region can include from about 10 nucleotides to about 10,000 nucleotides, from about 50 nucleotides to about 8000 nucleotides, from about 100 nucleotides to about 5000 nucleotides, from about 50 nucleotides to about 2000 nucleotides, from about 25 nucleotides to about 250 nucleotides, from about 50 nucleotides to about 200 nucleotides, or from about 75 nucleotides to about 150 nucleotides. For instance, classification region can be a differentially methylated region. “Differentially methylated region” or “DMR” refers to a region of DNA, such as a region of a genome, having a detectably different degree of methylation or a different methylation state in at least one cell or tissue type relative to the degree of methylation in the same region of DNA from at least one other cell or tissue type; or having a detectably different degree of methylation in at least one cell or tissue type obtained from a subject having a disease or disorder relative to the degree of methylation in the same region of DNA in the same cell or tissue type obtained from a healthy subject. In some embodiments, a differentially methylated region has a detectably higher degree of methylation in at least one cell or tissue type relative to the degree of methylation in the same region of DNA from at least one other cell or tissue type that contribute to cfDNA in healthy individuals, or from the same cell or tissue type from a healthy subject. In some embodiments, a differentially methylated region has a detectably lower degree of methylation in at least one cell or tissue type relative to the degree of methylation in the same region of DNA from at least one other cell or tissue type, such as other immune cell types and/or cell types that contribute to cfDNA in healthy individuals, or from the same cell or tissue type from a healthy subject.
[00107] Communications Network. As used herein, “communications network” refers to one or more portions of a network that may be an ad hoc network, an intranet, an extranet, a virtual private network (VPN), a local area network (LAN), a wireless LAN (WLAN), a wide area network (WAN), a wireless WAN (WWAN), a metropolitan area network (MAN), the Internet, a portion of the Internet, a portion of the Public Switched Telephone Network (PSTN), a plain old telephone service (POTS) network, a cellular telephone network, a wireless network, a Wi-Fi® network, another type of network, or a combination of two or more such networks. For example, a network or a portion of a network may include a wireless or cellular network and the coupling may be a Code Division Multiple Access (CDMA) connection, a Global System for Mobile communications (GSM) connection, or other type of cellular or wireless coupling. In this example, the coupling may implement any of a variety of types of data transfer technology, such as Single Carrier Radio Transmission Technology (1xRTT), Evolution- Data Optimized (EVDO) technology, General Packet Radio Service (GPRS) technology, Enhanced Data rates for GSM Evolution (EDGE) technology, third Generation Partnership Project (3GPP) including 3G, fourth generation wireless (4G) networks, Universal Mobile Telecommunications System (UMTS), High Speed Packet Access (HSPA), Worldwide Interoperability for Microwave Access (WiMAX), Long Term Evolution (LTE) standard, others defined by various standard setting organizations, other long range protocols, or other data transfer technology.
[00108] CpG: As used herein, “CpG” or “cytosine-guanine dinucleotide” refers to a cytosine-phosphate-guanine site within a nucleic acid molecule sequence such that a cytosine molecule is followed by a guanine molecule in a 5’ -> 3’ direction of the nucleic acid molecule sequence.
[00109] Deoxyribonucleic Acid or Ribonucleic Acid: As used herein, “deoxyribonucleic acid” or “DNA” refers to a natural or modified nucleotide which has a hydrogen group at the 2'-position of the sugar moiety. DNA can include a chain of nucleotides comprising four types of nucleotide bases: adenine (A), thymine (T), cytosine (C), and guanine (G). As used herein, “ribonucleic acid” or “RNA” refers to a natural or modified nucleotide which has a hydroxyl group at the 2’-position of the sugar moiety. RNA can include a chain of nucleotides comprising four types of nucleotides: A, uracil (U), G, and C. As used herein, the term “nucleotide” refers to a natural nucleotide or a modified nucleotide. Certain pairs of nucleotides specifically bind to one another in a complementary fashion (called complementary base pairing). In DNA, adenine (A) pairs with thymine (T) and cytosine (C) pairs with guanine (G). In RNA, adenine (A) pairs with uracil (U) and cytosine (C) pairs with guanine (G). When a first nucleic acid strand binds to a second nucleic acid strand made up of nucleotides that are complementary to those in the first strand, the two strands bind to form a double strand. As used herein, “nucleic acid sequencing data”, “nucleic acid sequencing information”, “sequence information”, “sequence representation”, “nucleic acid sequence”, “nucleotide sequence”, “genomic sequence”, “genetic sequence”, “fragment sequence”, “sequencing read”, or “nucleic acid sequencing read” denotes any information or data that is indicative of the order and identity of the nucleotide bases (e.g., adenine, guanine, cytosine, and thymine or uracil) in a molecule (e.g., a whole genome, whole transcriptome, exome, oligonucleotide, polynucleotide, or fragment) of a nucleic acid such as DNA or RNA. It should be understood that the present teachings contemplate sequence information obtained using all available varieties of techniques, platforms or technologies, including, but not limited to capillary electrophoresis, microarrays, ligation-based systems, polymerase-based systems, hybridization-based systems, direct or indirect nucleotide identification systems, pyrosequencing, ion- or pH-based detection systems, and electronic signature-based systems.
[00110] Differentially Methylated Region: As used herein, differentially methylated region" refers to a region of DNA, such as a region of a genome, having a detectably different degree of methylation or different methylation state in at least one cell or tissue type relative to the degree of methylation in the same region of DNA from at least one other cell or tissue type; or having a detectably different degree of methylation in at least one cell or tissue type obtained from a subject having a disease or disorder relative to the degree of methylation in the same region of DNA in the same cell or tissue type obtained from a healthy subject. In some embodiments, a differentially methylated region has a detectably higher degree of methylation (e.g., highly methylated region comprising 5-methylcytosines) in at least one cell or tissue type, such as at least one immune cell type, relative to the degree of methylation in the same region of DNA from at least one other cell or tissue type, such as other immune cell types and/or cell types that contribute to cfDNA in healthy individuals, or from the same cell or tissue type from a healthy subject. In one or more additional examples, a differentially methylated region can include a genomic region, such as a genomic, region corresponding to immune system function, that has a greater number of methylated nucleic acid molecules in a given sample due to a higher than expected turnover of cells related to the genomic region in an organ caused by the presence of a tumor in the organ. In some embodiments, a differentially methylated region has a detectably lower degree of methylation (e.g., unmethylated region comprising unmethylated cytosines) in at least one cell or tissue type, such as at least one immune cell type, relative to the degree of methylation in the same region of DNA from at least one other cell or tissue type, such as other immune cell types and/or cell types that contribute to cfDNA in healthy individuals, or from the same cell or tissue type from a healthy subject. [00111] Driver Mutation As used herein, “driver mutation” means a mutation that drives cancer progression.
[00112] Epigenetic Target Regions: As used herein, “epigenetic target regions” refers to target regions that may show sequence-independent differences in different cell or tissue types (e.g., different types of immune cells) or in neoplastic cells (e.g., tumor cells and cancer cells) relative to normal cells; or that may show sequence- independent differences (i.e., in which there is no change to the nucleotide sequence, e.g., differences in methylation, nucleosome distribution, or other epigenetic features) in DNA, such as cfDNA, from different cell types or from subjects having cancer relative to DNA, such as cfDNA, from healthy subjects, or in cfDNA originating from different cell or tissue types that ordinarily do not substantially contribute to cfDNA (e.g., immune, lung, colon, etc.) relative to background cfDNA (e.g., cfDNA that originated from hematopoietic cells). Examples of sequence-independent changes include, but are not limited to, changes in methylation (increases or decreases), nucleosome distribution, cfDNA fragmentation patterns, CCCTC-binding factor (“CTCF”) binding, transcription start sites (e.g., with respect to any one of more of binding of RNA polymerase components, binding of regulatory proteins, fragmentation characteristics, and nucleosomal distribution), and regulatory protein binding regions. Epigenetic target region sets thus include, but are not limited to, hypermethylation variable target region sets, hypomethylation variable target region sets, and fragmentation variable target region sets, such as CTCF binding sites and transcription start sites. For present purposes, loci susceptible to neoplasia-, tumor-, or cancer-associated focal amplifications and/or gene fusions may also be included in an epigenetic target region set because detection of a change in copy number by sequencing or a fused sequence that maps to more than one locus in a reference genome tends to be more similar to detection of exemplary epigenetic changes discussed above than detection of nucleotide substitutions, insertions, or deletions, e.g., in that the focal amplifications and/or gene fusions can be detected at a relatively shallow depth of sequencing because the^r detection does not depend on the accuracy of base calls at one or a few individual positions. An epigenetic target region set is a set of two or more epigenetic target regions. [00113] Immunotherapy As used herein, “immunotherapy” refers to treatment with one or more agents that act to stimulate the immune system so as to kill or at least to inhibit growth of cancer cells, and preferably to reduce further growth of the cancer, reduce the size of the cancer and/or eliminate the cancer. Some such agents bind to a target present on cancer cells; some bind to a target present on immune cells and not on cancer cells; some bind to a target present on both cancer cells and immune cells. Such agents include, but are not limited to, checkpoint inhibitors, genetically engineered immune cells and/or antibodies, including natural antibodies and genetically engineered antibodies. Checkpoint inhibitors are inhibitors of pathways of the immune system that maintain self-tolerance and modulate the duration and amplitude of physiological immune responses in peripheral tissues to minimize collateral tissue damage (see, e.g., Pardoll, Nature Reviews Cancer 12, 252-264 (2012)). Example agents include antibodies against any of PD-1 , PD-2, PD-L1 , PD-L2, CTLA-40, 0X40, B7.1 , B7He, LAG3, CD137, KIR, CCR5, CD27, or CD40. Other example agents include proinflammatory cytokines, such as IL-1 [3, IL-6, and TNF-a. Other example agents are T-cells activated against a tumor, such as T-cells designed to be activated by expressing a chimeric antigen receptor (CAR) targeting a tumor antigen and/or cell-surface protein engineered to be recognized by the T-cell.
[00114] Indel. As used herein, “indel” refers to a mutation that involves the insertion or deletion of nucleotides in the genome of a subject.
[00115] Machine-Readable Medium. As used herein, “machine-readable medium” refers to a component, device, or other tangible media able to store instructions and data temporarily or permanently and may include, but is not limited to, random-access memory (RAM), read-only memory (ROM), buffer memory, flash memory, optical media, magnetic media, cache memory, other types of storage (e.g., erasable programmable read-only memory (EEPROM)) and/or any suitable combination thereof. The term "machine- readable medium" may be taken to include a single medium or multiple media (e.g., a centralized or distributed database, or associated caches and servers) able to store instructions. The term "machine-readable medium" shall also be taken to include any medium, or combination of multiple media, that is capable of storing instructions (e.g., code) for execution by a machine and that when executed by one or more processors of the machine, cause the machine to perform any one or more of the methodologies described herein. Accordingly, a "machine-readable medium" refers to a single storage apparatus or device, as well as "cloud-based" storage systems or storage networks that include multiple storage apparatus or devices. The term "machine-readable medium" excludes signals per se.
[00116] Maximum MAF: As used herein, “maximum MAF” or “max MAF” refers to the maximum MAF (mutant allele fraction) of all somatic tumor variants in a sample.
[00117] Methylation: As used herein, “methylation” or “DNA methylation” refers to addition of a methyl group to a nucleotide base in a nucleic acid molecule. In some embodiments, methylation refers to addition of a methyl group to a cytosine at a CpG site. In some embodiments, DNA methylation refers to addition of a methyl group to adenine, such as in N6-methyladenine. In some embodiments, DNA methylation is 5-methylation (modification of the 5th carbon of the 6-carbon ring of cytosine). In some embodiments, 5-methylation refers to addition of a methyl group to the 5C position of the cytosine to create 5-methylcytosine (5mC). In some embodiments, methylation comprises a derivative of 5mC. Derivatives of 5mC include, but are not limited to, 5- hydroxymethylcytosine (5-hmC), 5-formylcytosine (5-fC), and 5-caryboxylcytosine (5- caC). In some embodiments, DNA methylation is 3C methylation (modification of the 3rd carbon of the 6-carbon ring of cytosine). In some embodiments, 3C methylation comprises addition of a methyl group to the 3C position of the cytosine to generate 3- methylcytosine (3mC). Methylation can also occur at non CpG sites, for example, methylation can occur at a CpA, CpT, or CpC site. DNA methylation can change the activity of methylated DNA region. For example, when DNA in a promoter region is methylated, transcription of the gene may be repressed. DNA methylation is critical for normal development and abnormality in methylation may disrupt epigenetic regulation. The disruption, e.g., repression, in epigenetic regulation may cause diseases, such as cancer. Promoter methylation in DNA may be indicative of cancer.
[00118] Methylation rate: As used herein, “methylation rate” refers to the probability, likelihood, or percentage that a given base (for example: cytosine residue in a CpG) is methylated on a DNA molecule at a particular genomic region analyzed in the sample. In some embodiments, the methylation rate may be applied to a defined region that comprises one or more potentially methylated bases. In some embodiments, the methylation rate refers to the percentage of CpG residues methylated in a DNA molecule. In some embodiments, the methylation rate refers to the percentage of CpG residues methylated in molecules aligned to particular genomic position or genomic region. Methylation rate can be measured by a variety of methods including, but not limited to, either using bisulfite sequencing (any single base resolution like TAPS, EM-SEQ, etc.) or using partitioning (DNA molecule resolution), such as DNA methylation partitioning using methyl-binding antibodies or proteins. Methylation rate can be measured in different ways. One estimation can be by counting how many DNA fragments end up in each methylation dependent partition or by counting the number of converted CpGs per fragment in the case of bisulfite sequencing or any other base-level resolution sequencing methods, such as qPCR-based methods that can use either converted DNA or methyl- precipitated DNA. In addition, in the case of methylation dependent partitioning, the rate calculation can be normalized using a set of predefined genomic control regions with known methylation state (i.e., positive control regions and/or negative control regions) or spiked- in synthetic DNA with known methylation state, deriving rate-parametrized partition distributions and estimating the rate using a maximum likelihood approach. In various examples, methylation rate can be calculated by dividing or “normalizing” the count of methylated molecules corresponding to one or more genomic regions by the number of molecules present within the genomic control regions. In one or more examples, the methylation rate can be determined by measuring an abundance of sequencing reads that correspond to a portion of a genomic region. The portion of the genomic region can include a number of genomic locations of the genomic region for which at least a threshold number of sequencing reads overlap.
[00119] Methylation Status: As used herein, “methylation status” or “methylation state” can refer to the presence or absence of methyl group on a DNA base (e.g., cytosine) at a particular genomic position in a nucleic acid molecule. It can also refer to the degree of methylation in a nucleic acid sequence (e.g., highly methylated, low methylated, intermediately methylated or unmethylated nucleic acid molecules). The methylation status can also refer to the number of nucleotides methylated in a particular nucleic acid molecule. [00120] Mutant Allele Fraction As used herein, “mutant allele fraction”, “mutation dose,” or “MAF” refers to the fraction of nucleic acid molecules harboring an allelic alteration or mutation at a given genomic position in a given sample. MAF is generally expressed as a fraction or a percentage. For example, an MAF can be less than about 0.5, 0.1 , 0.05, or 0.01 (i.e., less than about 50%, 10%, 5%, or 1 %) of all somatic variants or alleles present at a given locus.
[00121] Mutation. As used herein, “mutation” refers to a variation from a known reference sequence and includes mutations such as, for example, single nucleotide variants (SNVs), copy number variants or variations (CNVs)Zaberrations, insertions or deletions (indels), gene fusions, transversions, translocations, frame shifts, duplications, repeat expansions, and epigenetic variants.. A mutation can be a germline or somatic mutation. In some examples, a reference sequence for purposes of comparison is a wildtype genomic sequence of the species of the subject providing a test sample, typically the human genome.
[00122] Mutation Count. As used herein, “mutation count” or “mutational count” refers to the number of somatic mutations in a whole genome or exome or targeted regions of a nucleic acid sample.
[00123] Neoplasm. As used herein, the terms “neoplasm” and “tumor” are used interchangeably. They refer to abnormal growth of cells in a subject. A neoplasm or tumor can be benign, potentially malignant, or malignant. A malignant tumor is referred to as a cancer or a cancerous tumor.
[00124] Next Generation Sequencing: As used herein, “next generation sequencing” or “NGS” refers to sequencing technologies having increased throughput as compared to traditional Sanger- and capillary electrophoresis-based approaches, for example, with the ability to generate hundreds of thousands of relatively small sequencing reads at a time. Some examples of next generation sequencing techniques include, but are not limited to, sequencing by synthesis, sequencing by ligation, and sequencing by hybridization.
[00125] Nucleic Acid Tag: As used herein, “nucleic acid tag” refers to a short nucleic acid (e.g., less than about 500 nucleotides, about 100 nucleotides, about 50 nucleotides, or about 10 nucleotides in length), used to distinguish nucleic acids from different samples (e g., representing a sample index), or different nucleic acid molecules in the same sample (e.g., representing a molecular barcode), of different types, or which have undergone different processing. The nucleic acid tag comprises a predetermined, fixed, non-random, random or semi-random oligonucleotide sequence. Such nucleic acid tags may be used to label different nucleic acid molecules or different nucleic acid samples or sub-samples. Nucleic acid tags can be single-stranded, double-stranded, or at least partially double-stranded. Nucleic acid tags optionally have the same length or varied lengths. Nucleic acid tags can also include double-stranded molecules having one or more blunt-ends, include 5’ or 3’ single-stranded regions (e.g., an overhang), and/or include one or more other single-stranded regions at other locations within a given molecule. Nucleic acid tags can be attached to one end or to both ends of the other nucleic acids (e.g., sample nucleic acids to be amplified and/or sequenced). Nucleic acid tags can be decoded to reveal information such as the sample of origin, form, or processing of a given nucleic acid. For example, nucleic acid tags can also be used to enable pooling and/or parallel processing of multiple samples comprising nucleic acids bearing different molecular barcodes and/or sample indexes in which the nucleic acids are subsequently being deconvolved by detecting (e.g., reading) the nucleic acid tags. Nucleic acid tags can also be referred to as identifiers (e.g., molecular identifier, sample identifier). Additionally, or alternatively, nucleic acid tags can be used as molecular identifiers (e.g., to distinguish between different molecules or amplicons of different parent molecules in the same sample or sub-sample). This includes, for example, uniquely tagging different nucleic acid molecules in a given sample, or non-uniquely tagging such molecules. In the case of non-unique tagging applications, a limited number of tags (i.e., molecular barcodes) may be used to tag each nucleic acid molecule such that different molecules can be distinguished based on their endogenous sequence information (for example, start and/or stop positions where they map to a selected reference sequence, a sub-sequence of one or both ends of a sequence, and/or length of a sequence) in combination with at least one molecular barcode. A sufficient number of different molecular barcodes are used such that there is a low probability (e.g., less than about a 10%, less than about a 5%, less than about a 1 %, or less than about a 0.1 % chance) that any two molecules may have the same endogenous sequence information (e.g., start and/or stop positions, subsequences of one or both ends of a sequence, and/or lengths) and also have the same molecular barcode.
[00126] Polynucleotide: As used herein, “polynucleotide”, “nucleic acid”, “nucleic acid molecule”, “polynucleotide molecule”, or “oligonucleotide” refers to a linear polymer of nucleosides (including deoxyribonucleosides, ribonucleosides, or analogs thereof) joined by internucleosidic linkages. A polynucleotide can comprise at least three nucleosides. Oligonucleotides often range in size from a few monomeric units, e.g., 3-4, to hundreds of monomeric units. Whenever a polynucleotide is represented by a sequence of letters, such as “AGCTG,” it will be understood that the nucleotides are in 5’ -> 3’ order from left to right and that in the case of DNA, “A” denotes deoxyadenosine, “C” denotes deoxycytidine, “G” denotes deoxyguanosine, and “T” denotes deoxythymidine, unless otherwise noted. The letters A, C, G, and T may be used to refer to the bases themselves, to nucleosides, or to nucleotides comprising the bases, as is standard in the art.
[00127] Probe As used herein, “probe” refers to a polynucleotide comprising a functionality. The functionality can be a detectable label (fluorescent), a binding moiety (biotin), or a solid support (a magnetically attractable particle or a chip). Probes can include single-stranded DNA/RNA polynucleotides or double stranded DNA polynucleotides that hybridize to target nucleic acid sequences (e.g., SureSelect® probes, Agilent Technologies). Sequence capture using probes generally depends, in part, on the number of consecutive nucleotides in at least a portion of the target nucleic acid sequence that is complementary (or nearly complementary) to the sequence of the probe. In some examples, probes can correspond to driver mutations.
[00128] Processing: As used herein, the terms “processing”, “calculating”, and “comparing” can be used interchangeably. In certain applications, the terms refer to determining a difference, e.g., a difference in number or sequence. For example, gene expression, copy number variation (CNV), indel, and/or single nucleotide variant (SNV) values or sequences can be processed.
[00129] Processor. As used herein, “processor” refers to any circuit or virtual circuit (a physical circuit emulated by logic executing on an actual processor) that manipulates data values according to control signals (e.g., "commands," "op codes," "machine code," etc.) and which produces corresponding output signals that are applied to operate a machine. A processor may, for example, be a CPU, a RISC processor, a CISC processor, a GPU, a DSP, an ASIC, a RFIC or any combination thereof. A processor may further be a multi-core processor having two or more independent processors (sometimes referred to as "cores") that may execute instructions contemporaneously.
[00130] Promoter Region As used herein, “promoter region” refers to a DNA sequence recognized by the synthetic machinery of the cell, or introduced synthetic machinery, required to initiate the specific transcription of a gene.
[00131] Quantitative Measures: As used herein, “quantitative measures” refers to an absolute or relative measure. A quantitative measure can be, without limitation, a number, a statistical measurement (e.g., frequency, mean, median, standard deviation, or quantile), or a degree or a relative quantity (e.g., high, medium, and low). A quantitative measure can be a ratio of two quantitative measures. A quantitative measure can be a linear combination of quantitative measures. A quantitative measure may be a normalized measure.
[00132] Reference Sequence: As used herein, “reference sequence” refers to a known sequence used for purposes of comparison with experimentally determined sequences. For example, a known sequence can be an entire genome, a chromosome, or any segment thereof. A reference sequence can include at least about 20, at least about 50, at least about 100, at least about 200, at least about 250, at least about 300, at least about 350, at least about 400, at least about 450, at least about 500, at least about 1000, or more nucleotides. A reference sequence can align with a single contiguous sequence of a genome or chromosome or can include non-contiguous segments that align with different regions of a genome or chromosome. Example reference sequences, include, for example, human genome reference sequences, such as, hG19 and hG38.
[00133] Sample As used herein, “sample” means anything capable of being analyzed by the methods and/or systems disclosed herein.
[00134] Sequencing: As used herein, “sequencing” refers to any of a number of technologies used to determine the sequence (e.g., the identity and order of monomer units) of a biomolecule, e.g., a nucleic acid such as DNA or RNA. Example sequencing methods include, but are not limited to, targeted sequencing, single molecule real-time sequencing, exon or exome sequencing, intron sequencing, electron microscopy-based sequencing, panel sequencing, transistor-mediated sequencing, direct sequencing, random shotgun sequencing, Sanger dideoxy termination sequencing, whole-genome sequencing, sequencing by hybridization, pyrosequencing, capillary electrophoresis, duplex sequencing, cycle sequencing, single-base extension sequencing, solid-phase sequencing, high-throughput sequencing, massively parallel signature sequencing, emulsion PCR, co-amplification at lower denaturation temperature-PCR (COLD-PCR), multiplex PCR, sequencing by reversible dye terminator, paired-end sequencing, near- term sequencing, exonuclease sequencing, sequencing by ligation, short-read sequencing, single-molecule sequencing, sequencing-by-synthesis, real-time sequencing, reverse-terminator sequencing, nanopore sequencing, 454 sequencing, Solexa Genome Analyzer sequencing, SOLiD™ sequencing, MS-PET sequencing, and a combination thereof. In some implementations, sequencing can be performer by a gene analyzer such as, for example, gene analyzers commercially available from Illumina, Inc., Pacific Biosciences, Inc., or Applied B iosystem s/Thermo Fisher Scientific, among many others.
[00135] Single Nucleotide Variant As used herein, “single nucleotide variant” or “SNV” means a mutation or variation in a single nucleotide that occurs at a specific position in the genome.
[00136] Somatic Mutation As used herein, “somatic mutation” means a mutation in the genome that occurs after conception. Somatic mutations can occur in any cell of the body except germ cells and accordingly, are not passed on to progeny.
[00137] Specifically binds: As used herein, “specifically binds” in the context of a probe or other oligonucleotide and a target sequence means that under appropriate hybridization conditions, the oligonucleotide or probe hybridizes to its target sequence, or replicates thereof, to form a stable probe:target hybrid, while at the same time formation of stable probe: non-target hybrids is minimized. Thus, a probe hybridizes to a target sequence or replicate thereof to a sufficiently greater extent than to a non-target sequence, to enable capture or detection of the target sequence. Appropriate hybridization conditions are well-known in the art, may be predicted based on sequence composition, or can be determined by using routine testing methods (see, e.g., Sambrook et al., Molecular Cloning, A Laboratory Manual, 2nd ed. (Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 1989) at §§ 1.90-1.91 , 7.37-7.57, 9.47-9.51 and 11.47- 11.57, particularly §§ 9.50-9.51 , 11.12-11.13, 11.45-11.47 and 11.55-11.57, incorporated by reference herein).
[00138] Subject. As used herein, “subject” refers to an animal, such as a mammalian species (e.g., human) or avian (e.g., bird) species, or other organism, such as a plant. More specifically, a subject can be a vertebrate, e.g., a mammal such as a mouse, a primate, a simian or a human. Animals include farm animals (e.g., production cattle, dairy cattle, poultry, horses, pigs, and the like), sport animals, and companion animals (e.g., pets or support animals). A subject can be a healthy individual, an individual that has or is suspected of having a disease or a predisposition to the disease, or an individual that is in need of therapy or suspected of needing therapy. The terms “individual” or “patient” are intended to be interchangeable with “subject.”
[00139] For example, a subject can be an individual who has been diagnosed with having a cancer, is going to receive a cancer therapy, and/or has received at least one cancer therapy. The subject can be in remission of a cancer. As another example, the subject can be an individual who is diagnosed of having an autoimmune disease. As another example, the subject can be a female individual who is pregnant or who is planning on getting pregnant, who may have been diagnosed of or suspected of having a disease, e.g., a cancer, an auto-immune disease.
[00140] Target Region. As used herein, “target region” refers to a genomic locus targeted for identification and/or capture, for example, by using probes (e.g., through sequence complementarity). A "target region set” or “set of target regions” refers to a plurality of genomic loci targeted for identification and/or capture, for example, by using a set of probes (e.g., through sequence complementarity) or by using a set of PGR primers specific to the plurality of genomic loci.
[00141] Threshold. As used herein, “threshold” refers to a predetermined value used to characterize experimentally determined values of the same parameter for different samples depending on their relation to the threshold.
[00142] Tumor Fraction. As used herein, “tumor fraction” refers to the estimate of the fraction of nucleic acid molecules derived from a tumor in a given sample. For example, the tumor fraction of a sample can be a measure derived from the max MAF of the sample or pattern of sequencing coverage of the sample or length of the cfDNA fragments in the sample or any other selected feature of the sample. In some instances, the tumor fraction of a sample is equal to the max MAF of the sample.
[00143] Variant: As used herein, a “variant” can be referred to as an allele. A variant is usually presented at a frequency of 50% (0.5) or 100% (1 ), depending on whether the allele is heterozygous or homozygous. For example, germline variants are inherited and usually have a frequency of 0.5 or 1. Somatic variants; however, are acquired variants and usually have a frequency of < 0.5. Major and minor alleles of a genetic locus refer to nucleic acids harboring the locus in which the locus is occupied by a nucleotide of a reference sequence, and a variant nucleotide different than the reference sequence respectively. Measurements at a locus can take the form of allelic fractions (AFs), which measure the frequency with which an allele is observed in a sample.
DETAILED DESCRIPTION
[00144] Cancer is usually caused by the accumulation of mutations within genes of an individual's cells, at least some of which result in improperly regulated cell division. Such mutations can include single nucleotide variations (SNVs), gene fusions, insertions, transversions, translocations, and inversions. These mutations can also include copy number variations that correspond to an increase or a decrease in the number of copies of a gene within a tumor genome relative to an individual’s noncancerous cells. An extent of mutations present in cell-free nucleic acids and an amount of mutated cell-free nucleic acids of a sample can be used as biomarkers to determine tumor progression, predict patient outcome, and refine treatment choices. In various examples, the extent of mutations present in cell-free nucleic acids can be indicated by tumor cells copy number and tumor fraction for a given sample.
[00145] Additionally, cancer can be indicated by non-sequence modifications, such as methylation. Examples of methylation changes in cancer include local gains of DNA methylation in the CpG islands at the TSS of genes involved in normal growth control, DNA repair, cell cycle regulation, and/or cell differentiation. This increased amount of methylation can be associated with an aberrant loss of transcriptional capacity of involved genes and occurs at least as frequently as point mutations and deletions as a cause of altered gene expression.
[00146] Thus, DNA methylation profiling can be used to detect aberrant methylation in DNA of a sample. The DNA can correspond to certain genomic regions (“differentially methylated regions” or “DMRs”) that are normally hypermethylated or hypomethylated in a given sample type (e.g., cfDNA from the bloodstream) but which may show an abnormal degree of methylation that correlates to a neoplasm or cancer, e.g., because of unusually increased contributions of tissues to the type of sample (e.g., due to increased shedding of DNA in or around the neoplasm or cancer) and/or from extents of methylation of the genome that are altered during development or that are perturbed by disease, for example, cancer or any cancer-associated disease.
[00147] The methods and systems described herein are directed to generating image data using data and techniques that are different from existing data and techniques and implementing a convolutional neural network architecture to analyze the image data. The convolutional neural network architecture can analyze the image data to determine one or more indications related to tumor-related biological conditions. For example, the image data can be analyzed by the convolutional neural network architecture to determine an indication of one or more cancer types being present or absent in subject. Additionally, the convolutional neural network architecture can analyze the image data to determine an indication of minimum residual disease. In still other examples, the convolutional neural network architecture can analyze the image data to determine responsiveness of subjects to one or more cancer treatments.
[00148] The image data analyzed by the convolutional neural network architecture is not generated from conventional sources of digital image data. For example, digital image data is typically generated using film or image sensors that process light entering through a lens of a camera. The image sensors can include pixels that are used to record amounts of energy from light that is incident on the pixels. Digital signal processing can then be used to generate images based on image sensor signals. Conventional image data can also be generated in medical contexts through the use of x-rays, contrast media, magnets, and radioactive materials to generate images of internal organs or other internal portions of subjects. To illustrate, medical imaging can implement techniques, such as ultrasound, magnetic resonance imaging, radiography, computed tomography, positron emission tomography, one or more combinations thereof, and the like.
[00149] In implementations described herein, image data is generated using sequencing data that includes sequence representations derived from nucleic acid molecules present in one or more samples obtained from subjects. In one or more examples, the image data can be generated from alphanumeric representations of sequences of nucleotides that correspond to nucleic acid molecules present in one or more samples. The image data can also be generated using numerical values of molecular characteristics of the nucleic acid molecules present in one or more samples. In various examples, pixels included in the image data can each comprise a pair of values. A first value of the pair of values of individual pixels can correspond to genomic positions of the nucleic acid molecules with respect to a reference sequence and a second value of the pair of values of individual pixels can correspond to values of one or more molecular characteristics of the nucleic acid molecules corresponding to the sequence representations. The one or more molecular characteristics can include at least one of length of individual sequence representations, number of cytosine-guanine dinucleotides present in individual sequence representations, number of methylated cytosine-guanine dinucleotides present in individual sequence representations, or number of restriction enzyme cut sites associated with the individual sequence representations.
[00150] The convolutional neural network architecture can include multiple convolutional neural networks with individual convolutional neural networks being implemented to analyze images generated from sequence representations that are aligned with an individual genomic region. The genomic regions being analyzed can correspond to genomic regions that are enriched in relation to one or more diagnostic assays. In situations where sequence representations that are aligned with 20 genomic regions are being analyzed, the convolutional neural network architecture can include 20 convolutional neural networks. In situations where sequence representations that are aligned with 200 genomic regions are being analyzed, the convolutional neural network architecture can include 200 convolutional neural networks. Additionally, in scenarios where sequence representations that are aligned with 2000 genomic regions are being analyzed, the convolutional neural network architecture can include 2000 convolutional neural networks. The output from the individual convolutional neural networks can then be aggregated and analyzed to determine an overall indication with respect to a tumor- related biological condition being present or absent in relation to one or more subjects.
[00151] By generating image data from unconventional sources, implementations described herein can leverage the use of sequence representation data obtained from patient samples rather than relying on data generated by cameras or medical imaging equipment. In addition, by providing image data based on sequencing data and/or molecular characteristic values to the convolutional neural networks, rather than sequencing data and/or molecular characteristic values themselves, implementations described herein are able to analyze image data that is more naturally suited to processing by the internal arrangement of layers of convolutional neural network. As a result, implementations described herein can provide more accurate results than in scenarios where sequencing data and/or molecular characteristic values are themselves analyzed by the convolutional neural networks. Further, the accuracy of indications of tumor-related biological conditions determined by the implementations is increased with respect to existing systems because information from many genomic regions can be analyzed using separate convolutional neural networks.
[00152] Figure 1 is a diagrammatic representation of an example computational architecture 100 that implements one or more convolutional neural networks to identify samples obtained from subjects in which a tumor-related biological condition is present, according to one or more example implementations. The tumor-related biological condition can correspond to one or more diseases. In one or more examples, the disease under consideration is a type of cancer. Non-limiting examples of such cancers include biliary tract cancer, bladder cancer, transitional cell carcinoma, urothelial carcinoma, brain cancer, gliomas, astrocytomas, breast carcinoma, metaplastic carcinoma, cervical cancer, cervical squamous cell carcinoma, rectal cancer, colorectal carcinoma, colon cancer, hereditary nonpolyposis colorectal cancer, colorectal adenocarcinomas, gastrointestinal stromal tumors (GISTs), endometrial carcinoma, endometrial stromal sarcomas, esophageal cancer, esophageal squamous cell carcinoma, esophageal adenocarcinoma, ocular melanoma, uveal melanoma, gallbladder carcinomas, gallbladder adenocarcinoma, renal cell carcinoma, clear cell renal cell carcinoma, transitional cell carcinoma, urothelial carcinomas, Wilms tumor, leukemia, acute lymphocytic leukemia (ALL), acute myeloid leukemia (AML), chronic lymphocytic (CLL), chronic myeloid (CML), chronic myelomonocytic (CMML), liver cancer, liver carcinoma, hepatoma, hepatocellular carcinoma, cholangiocarcinoma, hepatoblastoma, lung cancer, non-small cell lung cancer (NSCLC), mesothelioma, B-cell lymphomas, non-Hodgkin lymphoma, diffuse large B-cell lymphoma, Mantle cell lymphoma, T-cell lymphomas, nonHodgkin lymphoma, precursor T-lymphoblastic lymphoma/leukemia, peripheral T-cell lymphomas, multiple myeloma, nasopharyngeal carcinoma (NPC), neuroblastoma, oropharyngeal cancer, oral cavity squamous cell carcinomas, osteosarcoma, ovarian carcinoma, pancreatic cancer, pancreatic ductal adenocarcinoma, pseudopapillary neoplasms, acinar cell carcinomas, prostate cancer, prostate adenocarcinoma, skin cancer, melanoma, malignant melanoma, cutaneous melanoma, small intestine carcinomas, stomach cancer, gastric carcinoma, gastrointestinal stromal tumor (GIST), uterine cancer, or uterine sarcoma. In one or more additional examples, the tumor-related biological condition can include minimal residual disease (MRD). In still other examples, the tumor-related biological condition can correspond to one or more immunological disorders.
[00153] The example computational architecture 100 can include, at 102, determining sequence representations 104 corresponding to a reference sequence 106. In one or more examples, determining sequence representations 104 that correspond to the reference sequence 106 can include aligning the sequence representations 104 with the reference sequence 106. In one or more illustrative examples, individual sequence representations 104 can be aligned with the reference sequence 106 by determining an amount of homology between individual nucleotides of the sequence representations 104 and positions of the reference sequence 106. The amount of homology between a given sequence representation 104 and a portion of the reference sequence 106 can be determined using BLAST programs (basic local alignment search tools) and PowerBLAST programs (Altschul et al., J. Mol. Biol., 1990, 215, 403-410; Zhang and Madden, Genome Res., 1997, 7, 649-656) or by using the Gap program (Wisconsin Sequence Analysis Package, Genetics Computer Group, University Research Park, Madison Wis.), using default settings, which uses the algorithm of Needleman and Wunsch (J. Mol. Biol. 48; 443-453 (1970)). The amount of homology between a sequence representation and a portion of the reference sequence can also be determined using a Burrows-Wheeler aligner (Li, H., & Durbin, R. (2009). Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics, 25(14), 1754-1760).
[00154] The sequence representations 104 can include alphanumeric representations of nucleic acid molecules derived from one or more samples. For example, the sequence representations 104 can include, for individual nucleic acids, data that corresponds to a string of letters that represents the respective chains of nucleotides that correspond to the individual nucleic acid molecules derived from one or more samples.
[00155] The sequence representations 104 can be stored in one or more data files. To illustrate, the sequence representations 104 be stored in a FASTQ file that comprises a text-based sequencing data file format storing raw sequence data and quality scores. In one or more additional examples, the sequence representations 104 can be stored in a data file according to a binary base call (BCL) sequence file format. In one or more further examples, the sequence representations 104 can be stored in a BAM file. In one or more examples, the sequence representations 104 can comprise at least about one gigabyte (GB), at least about 2 GB, at least about 3GB, at least about 4 GB, at least about 5 GB, at least about 8 GB, or at least about 10 GB. An individual sequence representation included in the sequence representations 104 can be referred to herein as a “read” or a “sequencing read.” In various examples, individual nucleic acid molecules derived from one or more samples can correspond to multiple sequence representations 104 as a result of the amplification of the individual nucleic acid molecules derived from one or more samples. In situations where amplification is not performed with respect to nucleic acid molecules derived from one or more samples, individual nucleic acid molecules derived from the one or more samples can correspond to a single sequence representation as a result of the absence of amplification of the individual nucleic acid molecules.
[00156] When multiple sequence representations 104 correspond to a single nucleic acid molecule derived from one or more samples, a number of groups can be generated from the sequence representations 104 with each group corresponding to a single nucleic acid molecule derived from the one or more samples. In various examples, the groups of the sequence representations 104 that correspond to a single nucleic acid molecule can be referred to herein as “families.” In at least some examples, families of sequence representations 104 can correspond to a unique cell-free nucleic acid molecule present in a sample. In at least some examples, start and stop positions with respect to a reference sequence having a common molecular barcode can be used to determine groups of the sequence representations 104 that correspond to individual nucleic acid molecules. In one or more illustrative examples, an individual sequence representation 104 that represents a family of sequence representations and that corresponds to a single nucleic acid molecule derived from one or more samples can be referred to herein as a “consensus sequence representation.”
[00157] The one or more samples that are analyzed to derive the sequence representations 104 can include one or more biological fluid samples obtained from one or more subjects. In one or more examples, the one or more samples can include one or more blood-based sample obtained from one or more subjects. In one or more additional examples, the one or more samples can include one or more plasma samples obtained from the one or more subjects. In still other examples, the one or more samples can include one or more tissue samples obtained from one or more subjects. The one or more subjects providing the one or more samples analyzed to generate the sequence representations 104 can include one or more mammals. In one or more additional examples, the one or more subjects providing the one or more samples analyzed to generate the sequence representations 104 can include one or more humans. In one or more further illustrative examples, the one or more subjects providing the one or more samples analyzed to generate the sequence representations 104 can include one or more non-human mammals.
[00158] The sequence representations 104 derived from one or more samples can be aligned with a reference sequence 106. In various examples, the sequence representations 104 can be aligned with one or more genomic regions. In the illustrative example of Figure 1 , the sequence representations 104 are aligned with a genomic region 108. In one or more examples, the genomic region 108 can include a classification region that can include one or more mutations in subjects in which a tumor-related biological condition is present. In one or more additional examples, the genomic region 108 can include a target region that comprises a number of genomic locations that are enriched using probes of one or more diagnostic tests. The one or more diagnostic tests can include one or more assays performed to detect subjects in which one or more tumor- related biological conditions are present.
[00159] The sequence representations 104 can include a number of individual sequence representations including a first sequence representation 110, a second sequence representation 112, a third sequence representation 114, and a fourth sequence representation 116. In the illustrative example of Figure 1 , the first sequence representation 110, the second sequence representation 112, and the third sequence representation 114 are aligned with a portion of the genomic region 108 such that start positions and stop positions of the individual sequence representations 110, 112, 114 include or are within the start position and the stop position of the genomic region 108. In addition, a portion of the fourth sequence representation 116 is aligned with a portion of the genomic region 108 and an additional portion of the fourth sequence representation 116 is aligned with a portion of the reference sequence 106 that is outside of the genomic region 108. In the illustrative example of Figure 1 , the first sequence representation 110, the second sequence representation 112, and the third sequence representation 114 comprise aligned sequence representations 118 that comprise at least a subset of the sequence representations 104 having start positions and stop positions that include and/or are within the start position and the stop position of the genomic region 108.
[00160] The sequence representations 104 can be derived from sequencing data that is generated as part of one or more sequencing processes performed with respect to nucleic acids obtained from one or more samples. The sequencing data can include and/or be used to generate position data 120 that corresponds to the aligned sequence representations 118. The position data 120 can indicate start positions and stop positions for individual aligned sequence representations 118 in relation to positions of the reference sequence 106. In one or more additional examples, the position data 120 can indicate an offset of at least one of a start position or a stop position of the aligned sequence representations 118 in relation to at least one of a start position or a stop position of the genomic region 108. In one or more further examples, the position data 120 can indicate a chromosome that is corresponding to the aligned sequence representations 118.
[00161] Additionally, the computational architecture 100 can include, at 122, determining molecular characteristics 124 of the aligned sequence representations 118. The molecular characteristics 122 can be determined by analyzing sequencing data corresponding to the aligned sequence representations 118. In one or more examples, the sequencing data of the aligned sequence representations 118 can be analyzed to determine lengths of the individual aligned sequence representations 118. In one or more illustrative examples, the length of an aligned sequence representation 118 can correspond to a number of nucleotides included in the individual aligned sequence representation 118. Further, the sequencing data of the aligned sequence representations 118 can be analyzed to determine a number of cytosine-guanine dinucleotides (CpGs) included in the aligned sequence representations 118.
[00162] In various examples, the molecular characteristics 124 can be determined by analyzing methylation data corresponding to the aligned sequence representations 118. In at least some examples, the sequencing data can include methylation data. The methylation data can be determined by one or more nucleobase methylation state detection processes. In one or more examples, the methylation data can indicate modified nucleotides that include one or more methyl groups that are not present in unmodified forms of the nucleotides. In one or more illustrative examples, the methylation data can indicate modified cytosines. That is, in various examples, the methylation data can indicate positions of nucleic acid molecules derived from one or more samples where at least one of a 5-methylcytosine and/or a 5-hydroxymethylcytosine is located. For example, the methylation data can indicate discrete, individual positions of individual nucleic acid molecules derived from one or more samples that include at least one of a 5-methylcytosine and/or a 5-hydroxymethylcytosine. In one or more additional examples, the methylation data can indicate a group of positions of individual nucleic acid molecules derived from one or more samples that include at least one of a 5-methylcytosine and/or a 5-hydroxymethylcytosine.
[00163] In one or more additional examples, the methylation data can indicate a number of restriction enzyme cut sites for individual aligned sequence representations 118. In one or more further examples, the molecular characteristics 124 of the aligned sequence representations 118 can be determined by analyzing methylation data to determine a number of methylated cytosines in the individual aligned sequence representations 118 and/or a range of methylated cytosines in the individual aligned sequence representations 118.
[00164] Further, the computational architecture 100 can include, at 126, generating image data 128 based on the position data 120 and the molecular characteristics 124. The image data 128 can include a number of images, such as an example image 130. The images 130 can include a plurality of pixels with each pixel having a first value indicated by a position along the x-axis and a second value indicated by a position along the y-axis. The images 130 can have at least 500 pixels, at least 1000 pixels, at least 2000 pixels, at least 5000 pixels, at least 10,000 pixels, at least 25,000 pixels, at least 50,000 pixels, at least 75,000 pixels, at least 100,000 pixels, or more. In one or more illustrative examples, the first value of the pixels included in the images 130 can correspond to a genomic position. For example, the first values of the pixels in the images 130 along the x-axis can correspond to positions of the genomic region 108. In various examples, the first values of the pixels in the images 130 along the x-axis can correspond to individual positions of the genomic region 108. In one or more additional examples, the first values of the pixels of the images 130 along the x-axis can correspond to a group of positions of the genomic region 108. In these scenarios, the first values of the pixels of the images 130 along the x-axis can correspond to intervals of positions in the genomic region 108. In one or more illustrative examples, the interval corresponding to the first values of the pixels of the images 130 along the x-axis can correspond to every 2 positions of the genom ic region 108, every 3 positions of the genom ic region 108, every 4 positions of the genom ic region 108, every 5 positions of the genom ic region 108, every 6 positions of the genom ic region 108, every 7 positions of the genom ic region 108, every 8 positions of the genomic region 108, every 9 positions of the genomic region 108, every 10 positions of the genomic region 108, every 12 positions of the genomic region 108, every 14 positions of the genomic region 108, every 16 positions of the genomic region 108, every 18 positions of the genomic region 108, every 20 positions of the genomic region 108, every 25 positions of the genomic region 108, every 30 positions of the genomic region 108, every 40 positions of the genomic region 108, or every 50 positions of the genomic region 108.
[00165] The second values of the pixels along the y-axis can correspond to values of molecular characteristics of the aligned sequence representations 118. In one or more examples, the second values of the pixels of the images 130 along the y-axis can correspond to numbers of CpGs of the aligned sequence representations 118. In one or more additional examples, the second values of the pixels of the images 130 along the y- axis can correspond to a number of methylated CpGs of the aligned sequence representations 118. In one or more further examples, the second values of the pixels of the images 130 along the y-axis can correspond to a number of unmethylated CpGs of the aligned sequence representations 118. In still other examples, the second values of the pixels of the images 130 along the y-axis can correspond to restriction enzyme cut sites for the aligned sequence representations 118. In various examples, the second values of the pixels of the images 130 along the y-axis can correspond to lengths of the aligned sequence representations 118.
[00166] In one or more illustrative examples, the images 130 can include pixel values of the aligned sequence representations 118 such that individual pixels can be characterized by a pair of values. The pair of values includes the x-axis value for an individual aligned sequence representation 118 that varies with the genomic positions of the individual aligned sequence representation 118 in conjunction with the genomic region 108 and a constant y-axis value that corresponds to the value of a specified molecular characteristic 124 of the individual aligned sequence representation 118. For example, in scenarios where the molecular characteristic 124 being represented on the y-axis is length of sequence representations, an individual aligned sequence representation 118 having a length of 120 nucleotides can be represented on the x-axis by a series of genomic locations that are aligned with the genomic region 108 and on the y-axis by the value of 120. In an illustrative situation where the length is quantized according to intervals of 10 nucleotides, the individual aligned sequence representation 118 having a length of 120 nucleotides can be represented on the y-axis by the value of 12. In one or more additional examples, in situations where the molecular characteristic 124 being represented on the y-axis is the number of CpGs included in the aligned sequence representations 118, an individual aligned sequence representation 118 having 14 CpGs can be represented on the x-axis by a series of genomic locations that are aligned with the genomic region 108 and on the y-axis by the value of 14.
[00167] The pixels of the images 130 can also have an intensity. The intensity of pixels of the image 130 can correspond to a number of the aligned sequence representations 118 that correspond to a pixel having a respective x-value and a respective y-value. As the number of the aligned sequence representations 118 that correspond to a pixel increases, the intensity value for the pixel increases. In the illustrative example of Figure 1 , the intensity values of the pixels corresponding to a first line 132 have a first magnitude that corresponds to a first number of aligned sequence representations 118 having a specified value of a molecular characteristic 124 and a specified genomic location. Additionally, the intensity values of the pixels corresponding to a second line 134 have a second magnitude that is greater than the first magnitude and that corresponds to a second number of aligned sequence representations 118 that is greater than the first number and that have an additional value for a molecular characteristic 124 and an additional genomic location. Further, the intensity values of the pixels corresponding to a third line 136 have a third magnitude that is greater than the second magnitude and that corresponds to a third number of aligned sequence representations 118 that is greater than the second number and that have a further value for the molecular characteristic 124 and a further genomic location. In at least some examples, individual lines included in the image 130 can have comprise pixels having different intensity values such that some intensity values of the pixels of the line have one value that corresponds to one number of the aligned sequence representations 118 overlapping with a set of genomic locations and a value of a molecular characteristic 124 and that other intensity values of the pixels of the line have another value that corresponds to another number of the aligned sequence representations 118 overlapping an additional set of genomic locations and having the same value of the molecular characteristic 124.
[00168] The image data 128 can be provided to a computational system 138 that implements a convolutional neural network architecture 140. The convolutional neural network architecture 140 can include a number of convolutional neural networks. The individual convolutional neural networks can correspond to individual genomic regions. To illustrate, the convolutional neural network architecture 140 can include a first convolutional neural network that analyzes first image data generated in relation to a first genomic region and a second convolutional neural network that analyzes second image data generated in relation to a second genomic region.
[00169] The convolutional neural networks included in the convolutional neural network architecture 140 can include a number of convolutional layers with individual layers being implemented to apply one or more convolutional filters to portions of an image. The values of the pixels can be modified according to the values of the filter to generate output values for a convolutional layer that can correspond to an output image. In various examples, the convolutional neural networks included in the convolutional neural network architecture 140 can identify one or more features within the images included in the image data 128. For example, individual convolutional layers of the convolutional neural network architecture 140 can generate feature maps based on the data input to the individual convolutional layer. In various examples, the convolutional neural network architecture 140 can include one or more rectified linear unit (ReLLI) layers that apply one or more rectification functions to the feature maps generated by the convolutional layers.
[00170] In one or more examples, the convolutional neural networks of the convolutional neural network architecture 140 can include one or more pooling layers. To illustrate, the convolutional neural networks included in the convolutional neural network architecture 140 can include one or more Max Pooling layers. The pooling layers of the convolutional neural network architecture 140 can reduce a size of dimensions of the feature maps produced by the convolutional layers. That is, as data corresponding to an image moves through layers of the convolutional neural network architecture 140, the number of pixels used to represent the image is reduced. Subsequent to processing by the one or more pooling layers, a flattening process can be performed prior to passing the modified feature maps to the output layers. The flattening process can generate onedimensional vectors from two-dimensional image data. In one or more additional examples, the convolutional neural networks of the convolutional neural network architecture 140 can produce three-dimensional output that includes a depth or number of channels in addition to the two-dimensional output, such as x-values and y-values for pixels. In these scenarios, depth can correspond to the application of different learnable convolutional filters and the flattening process can include flattening of the depth dimension in addition to width and height dimensions.
[00171] In various examples, the convolutional neural networks of the convolutional neural network architecture 140 can include one or more output layers that correspond to one or more classifications. The one or more classifications can be indicated by one or more probabilities determined by the convolutional neural networks of the convolutional neural network architecture 140 for the one or more classifications. For example, the output layers of the convolutional neural networks of the convolutional neural network architecture 140 can include fully-connected layers. In at least some examples, the output layers of the convolutional neural networks of the convolutional neural network architecture 140 can include one or more Softmax layers that include a number of nodes that apply a Softmax activation function to generate output probabilities.
[00172] In one or more examples, the convolutional neural network architecture 140 can analyze the image data 128 that corresponds to a number of genomic regions to determine a tumor indication 142. In various examples, the tumor indication 142 can include at least one of a tumor fraction or a tumor burden. In addition, the tumor indication 142 can indicate a binary determination that a tumor is present in a subject or that a tumor is not present in a subject. Further, the tumor indication 142 can indicate probabilities of a tumor being present in a subject. In still other examples the tumor indication 142 can correspond to an amount of progression of a tumor-related biological condition or an amount of regression of a tumor-related biological condition. In one or more additional examples, the tumor indication 142 can correspond to a responsiveness of a subject to one or more treatments. In one or more further examples, the tumor indication 142 can indicate a stage of cancer present in one or more subjects. In still other examples, the tumor indication 142 can correspond to one or more types of cancer present in the one or more subjects. In at least some examples, the convolutional neural network architecture 140 can implement a plurality of convolutional neural networks that individual generate a tumor indication that is combined into an overall tumor indication that is output by the convolutional neural network architecture 140 as the tumor indication 142. In one or more illustrative examples, the individual tumor indications produced by the plurality of convolutional neural networks can be combined using at least one of max pooling techniques, average pooling techniques, or one or more linear functions.
[00173] In various examples, the one or more nucleobase methylation state detection processes that are used to generate one or more of the molecular characteristics 124 can include one or more chemical processes and/or biochemical processes that impact a first type of nucleotide differently than a second type of nucleotide. For example, the one or more nucleobase methylation state detection processes implemented to generate one or more of the molecular characteristics 124 can include one or more reactions that cause at least one atomic and/or molecular moiety of the first type of nucleotide to be modified in a manner that is different from the manner in which the one or more reactions affect the second type of nucleotide. In one or more examples, the impact of the one or more nucleobase methylation state detection processes on a given type of nucleotide can be based on one or more previous modifications to the given type of nucleotide in relation to an unmodified form of the given type of nucleotide. That is, in various examples, a molecule corresponding to a given type of nucleotide may have been modified before being subjected to the one or more nucleobase methylation state detection processes. To illustrate, before being subjected to the one or more nucleobase methylation state detection processes, nucleotides of nucleic acid molecules derived from one or more samples can be modified due to mutations caused by the presence of a tumor in a subject. In at least some examples, the one or more nucleobase methylation state detection processes can modify the first type of nucleotide or the second type of nucleotide such that the nucleobase pairing of the first type of nucleotide or the second type of nucleotide is altered.
[00174] In one or more illustrative examples, the one or more nucleobase methylation state detection processes implemented to generate the molecular characteristics 124 can be performed on nucleic acid molecules included in one or more samples used to generate the number of sequence representations 104. The one or more nucleobase methylation state detection processes can modify a first type of nucleotide of the nucleic acid molecules in a first manner and one or more additional types of nucleotides of the nucleic acid molecules in a second manner. To illustrate, the one or more nucleobase methylation state detection processes can modify at least one of cytosines, guanines, thiamines, or adenines differently than at least one other of cytosines, guanines, thiamines, or adenines. In at least some examples, the one or more nucleobase methylation state detection processes can modify cytosines differently than guanines, thiamines, or adenines. In various examples, the one or more nucleobase methylation state detection processes can modify cytosines such that the modified cytosines no longer pair with guanines. For example, the one or more nucleobase methylation state detection processes can convert cytosines of the nucleic acid molecules included in one or more samples to uracils. In still other examples, the one or more nucleobase methylation state detection processes may not modify cytosines that were methylated prior to being subjected to the one or more nucleobase methylation state detection processes. In one or more examples, the one or more nucleobase methylation state detection processes may not modify 5-methylcytosines and/or 5- hydroxymethylcytosines of nucleic acid molecules derived from one or more samples In this way, the one or more nucleobase state detection processes can be used to differentiate cytosines that have been previously modified to include a 5-methyl group versus previously unmodified cytosines.
[00175] In one or more examples, the one or more nucleobase methylation state detection processes can include at least one of sodium bisulfite conversion and sequencing, Tet-assisted bisulfite sequencing (TAB-Seq), differential enzymatic cleavage, one or more single molecule sequencing methods, such as nanopore DNA sequencing, oxidative bisulfite (Ox-BS) conversion, APOBEC-coupled epigenetic (ACE) conversion, or direct methylation sequencing (DM-Seq).
[00176] In one or more additional examples, the one or more nucleobase methylation state detection processes can include one or more processes that separate nucleic acid molecules based on amounts of nucleotides of the nucleic acid molecules that have been previously modified. For example, the one or more nucleobase methylation state detection processes can determine a methylation rate for one or more regions of the nucleic acid molecules derived from one or more samples. In various examples, the one or more nucleobase methylation state detection processes can separate nucleic acid molecules included in one or more samples based on amounts of methylated cytosines included in CG regions of individual nucleic acid molecules. To illustrate, the one or more nucleobase methylation state detection processes can separate the nucleic acid molecules derived from one or more samples into a plurality of groups of nucleic acid molecules with individual groups of nucleic acid molecules corresponding to respective amounts of methylated cytosines of the nucleic acid molecules. The one or more nucleobase methylation state detection processes can include at least one of partitioning of nucleic acid molecules included in the one or more samples based on a strength of binding of the individual nucleic acid molecules to methyl binding domain (MBD) and, optionally, treatment with methylation sensitive restriction enzyme (MSRE) and/or methylation dependent restriction enzyme (MDRE). In various examples, a strength of binding of nucleic acid molecules to MBD can be determined by subjecting the nucleic acids to a series of washes having different concentrations of MBD. [00177] In at least some examples, the nucleobase methylation state detection processes can include one or more sequencing processes. For example, the nucleobase methylation state detection processes can include whole genome bisulfite sequencing, reduced representation bisulfite sequencing, targeted bisulfite sequencing, extended- representation bisulfite sequencing, or one or more combinations thereof. In one or more illustrative examples, whole genomic bisulfite sequencing can be performed according to the techniques described in T. Gong et al., “Analysis and performance assessment of the whole genome bisulfite sequencing data workflow: currently available tools and a practical guide to advance DNA methylation studies,” Small Methods, 6:e2101251 , 2022. In one or more additional illustrative examples, reduced representation bisulfite sequencing can be performed according to techniques described in Meissner, A., Gnirke, A., Bell, G.W., Ramsahoye, B., Lander, E.S., and Jaenisch, R. (2005). Reduced representation bisulfite sequencing for comparative high-resolution DNA methylation analysis. Nucleic acids research 33, 5868-5877. In one or more further illustrative examples, targeted bisulfite sequencing can be performed according to techniques described in D.A. Moser et al., “Targeted bisulfite sequencing: A novel tool for the assessment of DNA methylation with high sensitivity and increased coverage,” Psychoneuroendocrinology, 120:1 -8, 2020 and/or E. Leitao et al., “Locus-specific DNA methylation analysis by targeted deep bisulfite sequencing,” Methods Mol Biol, 1767:351-66, 2018. In still further illustrative examples, extended-representation bisulfite sequencing can be performed according to techniques described in Shareef, S.J., Bevill, S.M., Raman, A.T. et al. Extended- representation bisulfite sequencing of gene regulatory elements in multiplexed samples and single cells. Nat Biotechnol 39, 1086-1094 (2021 ).
[00178] Figure 2 is a diagrammatic representation of an example computational architecture 200 to generate image data derived from sample data and implementing a number of convolutional neural networks to analyze the image data for the detection of tumor-related biological conditions, according to one or more example implementations. At 202, the computational architecture 200 can include generating image data for individual genomic regions of a plurality of genomic regions. In this way, image data is generated for individual genomic regions. The individual genomic regions can include a first genomic region 204, a second genomic region 206, and a third genomic region 208. The genomic regions 204, 206, 208 can correspond to at least one of a classification region, a target region for a diagnostic assay, a control region for a diagnostic assay, or a differentially methylated region. In at least some examples, the genomic regions 204, 206, 208 can be included in a reference sequence.
[00179] Although the illustrative example of Figure 2 is directed to generating image data for three genomic regions, the computational architecture 200 can be implemented with respect to more genomic regions or fewer genomic regions. In various examples, the computational architecture 200 can be implemented in relation to hundreds of genomic regions, up to thousands of genomic regions. To illustrate, the computational architecture 200 can be implemented in relation to 5 genomic regions, 10 genomic regions, 25 genomic regions, 50 genomic regions, 100 genomic regions, 200 genomic regions, 400 genomic regions, 600 genomic regions, 800 genomic regions, 1000 genomic regions, 1500 genomic regions, 2000 genomic regions, 2500 genomic regions, 3000 genomic regions, 4000 genomic regions, 5000 genomic regions, or more.
[00180] In one or more examples, first sequence representations 210 can be produced with respect to the first genomic region 204, second sequence representations 212 can be produced with respect to the second genomic region 206, and third sequence representations 214 can be produced with respect to the third genomic region 208. The sequence representations 210, 212, 214 can correspond to nucleic acid molecules present in one or more samples obtained from one or more subjects. In at least some examples, the sequence representations 210, 212, 214 can correspond to nucleic acid molecules present in one or more samples provided by a subject that is being tested for a tumor-related biological condition. In one or more examples, the sequence representations 210, 212, 214 can include or correspond to sequencing reads produced in conjunction with one or more sequencing processes performed with respect to the one or more samples.
[00181] The first sequence representations 210 can have first molecule characteristics 216 and first position data 218, the second sequence representations 212 can have second molecular characteristics 220 and second position data 222, and the third sequence representations 214 can have third molecular characteristics 224 and third position data 226. The molecular characteristics 216, 220, 224 can include at least one of length of the respective sequence representations 210, 212, 214; number of CpGs present in the respective sequence representations 210, 212, 214; number of methylated CpGs present in the respective sequence representations 210, 212, 214; or number of restriction enzyme cut sites for the respective sequence representations 210, 212, 214. The position data 218, 222, 226 can indicate locations of the respective sequence representations 210, 212, 214 in relation to a reference genome. In one or more examples, the position data 218, 222, 226 can indicate at least one of start positions or stop positions of the respective sequence representations 210, 212, 214. Additionally, the position data 218, 222, 226 can indicate a chromosome corresponding to the respective sequence representations 210, 212, 214. Further, the position data 218, 222, 226 can indicate, for the respective sequence representations 210, 212, 214, an offset with respect to at least one of start positions of the genomic regions 204, 206, 208; stop positions of the genomic regions 204, 206, 208; start position of a chromosome; or a stop position of a chromosome.
[00182] The computational architecture 200 can generate first image data 228 corresponding to the first genomic region 204 based on the first molecular characteristics 216 and the first position data 218. The computational architecture 200 can also generate second image data 230 corresponding to the second genomic region 206 based on the second molecular characteristics 220 and the second position data 222. In addition, the computational architecture 200 can generate third image data 232 corresponding to the third genomic region 208 based on the third molecular characteristics 224 and the third position data 226. In various examples, one or more images can be generated for the individual genomic regions 204, 206, 208. In one or more examples, the image data 228, 230, 232 can include images having a number of pixels. Individual pixels can be represented by a pair of values. For example, individual pixels can be represented by first values corresponding to the position data 218, 222, 226 and second values corresponding to values of one or more molecular characteristics included in the molecular characteristics 216, 220, 224.
[00183] The individual pixels can also have intensity values that correspond to a number of the sequence representations 210, 212, 214 that overlap with the given position data and molecular characteristic values of the individual pixels. In one or more illustrative examples, the intensity values for individual pixels included in images of the first image data 228, the second image data 230, and the third image data 232 can correspond to normalized intensity values. To illustrate, the intensity values for pixel values of images included in the image data 228, 230, 232 can be normalized in relation to a maximum pixel intensity value for individual images included in the image data 228, 230, 232. In various examples, the intensity values for the pixels of images included in the image data 228, 230, 232 can include a logarithmic transformation of a normalized pixel value. In these scenarios, individual normalized pixel values can correspond to a number of sequence representations having the position value and molecular characteristic value for the pixel in relation to the number of sequence representations that correspond to a given genomic region. In at least some examples, the normalized pixel values can be determined in relation to quantitative measures, such as counts, of the number of sequence representations corresponding to a number of control regions. In one or more illustrative examples, a pixel of an image included in the first image data 228 can have a first value corresponding to a position of a nucleotide included in one or more of the first sequence representations 210 and a second value corresponding to a molecular characteristic of the one or more first sequence representations 210. Continuing with this example, the normalized intensity value for the pixel can be determined by a logarithmic transformation of a ratio of the number of first sequence representations 210 having the first value and the second value and the total number of the first sequence representations 210.
[00184] In various examples, the first image data 228 can include multiple images with individual images of the multiple images being generated based on different molecular characteristics. For example, the first image data 228 can include a first image that is generated using values of a first molecular characteristic and a second image that is generated using values of a second molecular characteristic. In these situations, the x- values of the pixels in the first image and the second image can be the same and the y- values for the pixels in the first image and the second image can be different. In one or more illustrative examples, the first image data 228 can include a first image that includes pixels with x-values based on the first position data 218 and y-values that correspond to lengths of the first sequence representations 210 and a second image that includes pixels with x-values based on the first position data 218 and y-values that correspond to numbers of CpGs of the first sequence representations 210. Additionally, the second image data 230 and the third image data 232 can include multiple images that are generated based on values of different molecular characteristics.
[00185] In various examples, where multiple images are generated for a sample using different molecular characteristics, the computational architecture 200 can include one or more convolutional neural networks dedicated to computationally analyzing images generated using a given molecular characteristic. For example, the computational architecture 200 can include one or more first convolutional neural networks that computationally analyze first images generated based on length of sequence representations and one or more second convolutional neural networks that computationally analyze second images generated based on number of methylated CpGs present in sequence representations. The outputs from the different molecular characteristic-based convolutional neural networks can be combined to generate a composite tumor indication. For example, in binary classification implementations, one or more first tumor indications produced by the one or more first convolutional neural networks corresponding to a first molecular characteristic can be combined using one or more pooling techniques with one or more second tumor indications produced by one or more second convolutional neural networks corresponding to a second molecular characteristic. In these scenarios, the pooling layer can be trained using binary loss. In scenarios where the convolutional neural networks of the computational architecture 200 are trained to produce outputs in relation to multiple cancer types, one or more first convolutional neural networks corresponding to a first molecular characteristic can generate multiple first outputs with each first output corresponding to an individual cancer type of a plurality of cancer types and one or more second convolutional neural networks corresponding to a second molecular characteristic can generate multiple second outputs with each second output corresponding to an individual cancer type of the plurality of cancer types. The first outputs and the second outputs can be combined using one or more pooling techniques with training of the computational architecture 200 being performed using one or more softmax techniques.
[00186] The first image data 228, the second image data 230, and the third image data 232 can be provided to a computational system 234. The computational system 234 can include multiple convolutional neural networks. The individual convolutional neural networks included in the computational system 234 can analyze one of the first image data 228, the second image data 230, or the third image data 232. For example, the computational system 234 can include a first region convolutional neural network 236, a second region convolutional neural network 238, and a third region convolutional neural network 240. The first region convolutional neural network 236 can analyze the first image data 228 to determine a first tumor indication 242 that corresponds to the first genomic region 204. Additionally, the second region convolutional neural network 238 can analyze the second image data 230 to determine a second tumor indication 244 that corresponds to the second genomic region 206. Further, the third region convolutional neural network 240 can analyze the third image data 232 to determine a third tumor indication 246. The tumor indications 242, 244, 246 can individually indicate at least one of a tumor fraction, a tumor burden, a probability of a tumor-related biological condition being present in a subject, a progression of a tumor-related biological condition, a regression of a tumor- related biological condition, presence or absence of a tumor-related biological condition, or responsiveness to treatment for a tumor-related biological condition. In one or more illustrative examples, the output of the individual region convolutional neural networks 236, 238, 240 can include a logit value that corresponds to the tumor indications 242, 244, 246.
[00187] At 248, the computational system 234 can determine an overall tumor indication based on individual tumor indications generated by the region convolutional neural networks included in the computational system 234. For example, the overall tumor indication can be a combination of the first tumor indication 242, the second tumor indication 244, and the third tumor indication 246. In at least some examples, the tumor indications 242, 244, 246 can be analyzed according to at least one of one or more machine learning techniques or one or more statistical techniques to determine the overall tumor indication. In various examples, the overall tumor indication can be determined by using a logistic regression technique in relation to the first tumor indication 242, the second tumor indication 244, and the third tumor indication 246. In still other examples, one or more non-linear computational methods can be applied to determine the overall tumor indication in relation to the first tumor indication 242, the second tumor indication 244, and the third tumor indication 246. To illustrate, at least one of one or more random forests methods or one or more boosting tree methods can be implemented to determine the overall tumor indication in relation to the first tumor indication 242, the second tumor indication 244, and the third tumor indication 246.
[00188] The region convolutional neural networks included in the computational system 234 can be trained individually. For example, first training data can be obtained for the first genomic region 204, second training data can be obtained for the second genomic region 206, and third training data can be obtained for the third genomic region. The first training data, the second training data, and the third training data can be generated from training samples obtained from first training subjects in which a tumor- related biological condition is present and from second training subjects in which a tumor- related biological condition is not present. In various examples, the first training data, the second training data, and the third training data can be labeled or unlabeled. In one or more illustrative examples, labeled training data can include a first indication for position data and molecular characteristic data corresponding to sequence representations produced in relation to samples obtained from first training subjects in which a tumor- related biological condition is present and a second indication for position data and molecular characteristic data corresponding to sequence representations produced in relation to samples obtained from second training subjects in which a tumor-related biological condition is not present.
[00189] In one or more examples, the first region convolutional neural network 236 can undergo a first training process based on the first training data corresponding to the first genomic region 204. Additionally, the second region convolutional neural network 238 can undergo a second training process based on the second training data corresponding to the second genomic region 206. Further, the third region convolutional neural network 240 can undergo a third training process based on the third training data. In one or more illustrative examples, the first training data, the second training data, and the third training data can include position data and molecular characteristic data for training sequence representations that correspond to the genomic regions 204, 206, 208 and that are generated in relation to the training samples. The training data can also include images generated from the position data and molecular characteristic data corresponding to the genomic regions 204, 206, 208 and obtained from the first training subjects and the second training subjects.
[00190] The training of the region convolutional neural networks 236, 238, 240 can include performing one or more feature extraction operations and one or more classification operations. Feature extraction operations can include identifying one or more variables and/or one or more sets of variables that can be used to make one or more predictions based on a set of input data. The feature extraction can determine relationships between one or more variables included in the training data and determine one or more measures of correlations between variables and/or groups of variables included in the training data. The classification operations can include classifying one or more pieces of information included in the training data according to one or more categories.
[00191] The feature extraction and classification operations can be used to determine one or more initial computational models for the region convolutional neural networks 236, 238, 240 that can make one or more determinations and/or predictions directed to tumor indications associated with the training data. Additionally, evaluation data can be used to evaluate the performance of the initial computational models produced by the feature extraction operations and the classification operations. The evaluation data can include a portion of the corpus of data that is different from the data used to train the region convolutional neural networks 236, 238, 240. In this way, the evaluation data that is used to evaluate the performance of the initial computational models is different from the training data that is used to generate the one or more initial computational models. Based on the performance of the initial computational models implemented with the evaluation data, the feature extraction and/or classification operations can be performed one or more additional times until the one or more computational models are produced. The one or more computational models can be produced after one or more iterations of feature extraction, classification, and prediction using the evaluation data. For example, a training process to generate the computational models of the region convolutional neural networks 236, 238, 240 can include at least 5 iterations, at least 10 iterations, at least 25 iterations, at least 50 iterations, at least 100 iterations, at least 500 iterations, at least 1000 iterations, or more In one or more implementations, the one or more computational models of the region convolutional neural networks 236, 238, 240 can be produced after the feature extraction, classification, and prediction operations produce computational models that satisfy one or more performance criteria, such as one or more convergence criteria or one or more accuracy criteria.
[00192] In various illustrative examples, the training process for the region convolutional neural networks 236, 238, 240 can include a forward phase and a backward phase. In the forward phase inputs pass through the layers of the region convolutional neural networks 236, 238, 240 to produce an output. In the backward phase, gradients based on the outputs from the forward phase are backpropagated and weights of the layers are modified. Data generated by one or more layers of the region convolutional neural networks 236, 238, 240 during the forward phase can be cached for later use in the backward phase of the training process. In one or more examples, the weights and/or biases of the individual layers of the region convolutional neural networks 236, 238, 240 can be determined using stochastic gradient descent techniques. In one or more additional examples, cross-entropy loss for classification probabilities generated by the region convolutional neural networks 236, 238, 240 during one or more iterations of the training process can be used to modify the weights and/or biases of the layers of the region convolutional neural networks 236, 238, 240. In at least some examples, weights corresponding to filters of one or more layers of the region convolutional neural networks 236, 238, 240 can be modified during a training process for the region convolutional neural networks 236, 238, 240.
[00193] Figure 3 is a diagrammatic representation of an example computational architecture 300 to implement one or more convolutional neural networks to detect a plurality of cancer types, according to one or more example implementations. The computational architecture 300 can include a computing system 302 that analyzes image data generated from sequencing data and molecular characteristics data to determine an indication of a cancer type present in one or more subjects. In various examples, the computing system 302 can perform at least a portion of the operations described with respect to Figure 1 and Figure 2. The computing system 302 can include one or more computing devices 304. The one or more computing devices 304 can include at least one of one or more desktop computing devices, one or more mobile computing devices, or one or more server computing device. In various examples, at least a portion of the one or more computing devices 304 can be included in a remote computing environment, such as a cloud computing environment. In one or more examples, the operations executed by the computing system 302 can be performed by, controlled by, and/or maintained by a single organization. In one or more additional examples, the operations executed by the computing system 302 can be performed by, controlled by, and/or maintained by multiple organizations.
[00194] The computing system 302 can include a number of region convolutional neural networks, such as an example region convolutional neural network 306. The region convolutional neural networks can analyze image data for a given genomic region to determine an indication of a tumor-related biological condition based on features of sequence representations aligned with the genomic region. In one or more examples, the region convolutional neural networks 306 can include or correspond to at least one of the convolutional neural network architecture 140 described in relation to Figure 1 or the region convolutional neural networks 236, 238, 240 described in relation to Figure 2. [00195] The region convolutional neural networks 306 can include a plurality of output layers with individua output layers of the region convolutional neural networks 306 corresponding to one or more specified cancer types. For example, the region convolutional neural networks 306 can perform first output computations 308 that produces a first cancer type output 310. The region convolutional neural networks 306 can also perform second output computations 312 that produces a second cancer type output 314. Additionally, the region convolutional neural networks 306 can perform third output computations 316 that generates a third cancer type output 318. In various examples, each output 310, 314, 318 can indicate a probability of a given cancer type being present in subjects. In one or more examples, the first output computations 308, the second output computations 312, and the third output computations 316 can be performed using one or more output layers of the region convolutional neural networks 306. In one or more illustrative examples, the first cancer type output 310, the second cancer type output 314, and the third cancer type output 318 can be produced using one or more softmax layers with each output 310, 314, 318 being determined using individual sets of weights corresponding to individual outputs 310, 314, 318.
[00196] The cancer type outputs 310, 314, 318 can correspond to an indication of different types of cancer being present in subjects based on data generated from sequence representations corresponding to a given genomic region. For example, first cancer type genomic region data 320 can be analyzed to determine the first cancer type output 310. In addition, second cancer type genomic region data 322 can be analyzed to determine the second cancer type output 314. Further, third cancer type genomic region data 324 can be analyzed to determine the third cancer type output 318. The genomic region data 320, 322, 324 can include at least one of sequencing data, methylation data, genomic position data, or molecular characteristics data that is produced by analyzing at least one of sequence representations or molecular data of nucleic acid molecules that align with a genomic region.
[00197] The region convolutional neural networks 306 can be trained using training data obtained from subjects in which the first cancer type, the second cancer type, and the third cancer type are present. In at least some examples, the training of the region convolutional neural networks 306 with respect to determining indications for multiple cancer types can be performed without using data obtained from subjects in which no cancer is detected. In still other examples, the training of the region convolutional neural networks 306 with respect to determining indications for multiple cancer types can be performing using data obtained from subjects in which no cancer is detected.
[00198] In one or more examples, the first output computations 308 can be performed using at least one of sequencing data, methylation data, genomic position data, or molecular characteristics data that is produced by analyzing at least one of sequence representations or molecular data of nucleic acid molecules that align with a genomic region and that is derived from samples obtained from subjects in which the first cancer type is present. Additionally, the second output computations 312 can be performed using at least one of sequencing data, methylation data, genomic position data, or molecular characteristics data that is produced by analyzing at least one of sequence representations or molecular data of nucleic acid molecules that align with the genomic region and that is derived from samples obtained from subjects in which the second cancer type is present. Further, the third output computations 316 can be performed using at least one of sequencing data, methylation data, genomic position data, or molecular characteristics data that is produced by analyzing at least one of sequence representations or molecular data of nucleic acid molecules that align with the genomic region and that is derived from samples obtained from subjects in which the third cancer type is present. In this way, the training of the region convolutional neural network 306 with respect to different cancer types can be performed using batches of training data corresponding to the individual cancer types. The training process to perform the output computations 308, 312, 316 of the region convolutional neural networks 306 can correspond to the training process for the region convolutional neural networks 236, 238, 240 included in the computational system 234 described in relation to Figure 2.
[00199] In one or more examples, the first cancer type output 310 can correspond to a probability of the first type of cancer being present in one or more subjects. In one or more additional examples, the second cancer type output 314 can correspond to a probability of the second type of cancer being present in one or more subjects. In one or more further examples, the third cancer type output 318 can correspond to a probability of the second type of cancer being present in one or more subjects. In one or more illustrative examples, the first cancer type output 310, the second cancer type output 314, and the third cancer type output 318 can include individual logit values corresponding to the first cancer type, the second cancer type, and the third cancer type.
[00200] The computing system 302 can, at 326, analyze the cancer type outputs to determine a final cancer type output 328 for the genomic region associated with the region convolutional neural network 306. For example, the computational system 302 can analyze the first cancer type output 310, the second cancer type output 314, and the third cancer type output 318 to determine the final cancer type output 328. In at least some examples, the computing system 302 can implement one or more argmax functions based on the first cancer type output 310, the second cancer type output 314, and the third cancer type output 318 to determine the final cancer type output 328.
[00201] In one or more examples, the computing system 302 can determine an output of the region convolutional neural network 306 having a highest value. To illustrate, the computing system 302 can analyze the first cancer type output 310, the second cancer type output 314, and the third cancer type output 318 to determine the output 310, 314, 318 having a highest probability value. In one or more additional examples, the computing system 302 can analyze the output values of the region convolutional neural network 306 in relation to one or more threshold values. The one or more threshold values can correspond to a likelihood of a one or more types of cancer being present in a subject. In various examples, the computing system 302 can determine at least one of the first cancer type output 310, the second cancer type output 314, or the third cancer type output 318 having at least a threshold value. In still other examples, the computing system 302 can determine at least one of the first cancer type output 310, the second cancer type output 314, or the third cancer type output 318 having no greater than a threshold value. In one or more illustrative examples, the computing system 302 can determine at least one of the first cancer type output 310, the second cancer type output 314, or the third cancer type output 318 corresponding to a minimum probability of the respective type of cancer being present in one or more subjects.
[00202] The final cancer type output 328 can correspond directly to an output generated by an output layer 308, 312, 316 that is identified by the computing system 302. In one or more examples, the computing system 302 can determine that from among the first cancer type output 310, the second cancer type output 314, and the third cancer type output 318, the first cancer type output 310 corresponds to one or more criteria. In these scenarios, the final cancer type output 328 can include or correspond to the first cancer type output 310. In one or more additional examples, the computing system 302 can analyze and/or modify at least one of the cancer type outputs 310, 314, 318 to determine the final cancer type output 328. To illustrate, the final cancer type output 328 produced by the computing system 302 can include at least one of a tumor fraction, a tumor burden, an indication of tissue of origin, a probability of at least one type of cancer being present in one or more subjects, an indication of the presence of at least one type of cancer, an indication of the absence of at least one type of cancer, an indication of progression of at least one type of cancer, an indication of regression of at least one type of cancer, or an indication of responsiveness of one or more subjects to a treatment provided in relation to at least one type of cancer.
[00203] Although the illustrative example of Figure 3 shows a single region convolutional neural network 306 with respect to the computing system 302, the computing system 302 can include multiple region convolutional neural networks 306 with the individual region convolutional neural networks that analyze at least one of sequencing data, methylation data, genomic position data, or molecular characteristics data corresponding to at least one of sequence representations or nucleic acid molecules to determine indications related to multiple types of cancer being present in one or more subjects. In various examples, the individual region convolutional neural networks of a plurality of region convolutional neural networks of the computing system 302 can include a same number of output layers corresponding to different cancer types for the individual genomic regions associated with the respective region convolutional neural networks. In one or more additional examples, at least a portion of the individual region convolutional neural networks of the computing system 302 can include a different number of output layers corresponding to different cancer types for the individual genomic regions associated with the respective region convolutional neural networks. In at least some examples, the computing system 302 can aggregate the cancer type outputs from the individual region convolutional neural networks to determine an overall cancer type output for one or more subjects. In one or more illustrative examples, at least one of one or more machine learning models or one or more statistical techniques can be implemented to determine an overall cancer type output for one or more subjects based on final cancer type outputs 328 from a plurality of region convolutional neural networks corresponding to different genomic regions. To illustrate, a logistic regression model can be implemented by the computing system 302 to aggregate final cancer type outputs 328 from a plurality of region convolutional neural networks to determine an overall cancer type output for one or more subjects.
[00204] Figure 4 is a flow diagram of an example process 400 to generate image data and implement one or more convolutional neural networks to analyze the image data for the detection of a tumor-derived biological condition, according to one or more implementations. At 402, the process 400 can include obtaining sequencing data indicating a number of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects. In one or more examples, the sequencing data can include sequencing reads generated by one or more sequencing operations performed with respect to the nucleic acid molecules. In various examples, the sequencing data can also include methylation data indication one or more methylated CpGs present in the nucleic acid molecules.
[00205] In addition, the process 400 can include, at 404, determining, based on the sequencing data, a group of sequence representations that are aligned with respect to one or more portions of a genomic region. The genomic region can be included in a number of genomic regions that are enriched as part of a diagnostic test to determine the presence of tumors in subjects.
[00206] The process 400 can also include, at 406, determining, based on the group of sequence representations, values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations. The one or more molecular characteristics can include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the group of sequence representations. The one or more molecular characteristics can also include a number of methylated cytosine-guanine dinucleotides present in an individual sequence representation of a group of sequence representations. In addition, the one or more molecular characteristics can include a length of the individual sequence representations of the group of sequence representations. Further, the one or more molecular characteristics can include a number of restriction enzyme cut sites in the individual sequence representations of the group of sequence representations. in still other examples, the one or more molecular characteristics can include a number of sequence representations derived from a sample after one or more sequencing operations are performed that correspond to a given nucleic acid included in the sample. [00207] Further, at 408, the process 400 can include generating, based on the group of sequence representations, one or more images that include a plurality of pixels. Individual pixels of the plurality of pixels can comprise a first value that corresponds to a genomic location within the genomic region and a second value that corresponds to the one or more molecular characteristics. The genomic locations that correspond to the first values of the plurality of pixels can correspond to an interval that comprises a plurality of nucleotides. Additionally, individual pixels of the plurality of pixels can include an intensity value indicating a number of the sequence representations included in the group of sequence representations having the first value and the second value. Intensity values of the plurality of pixels can increase as the number of sequence representation having the first value and the second value increases. Further, the intensity values of the plurality of pixels can be normalized based on a maximum intensity value for the plurality of pixels. In one or more illustrative examples, intensity value for the individual pixels of the plurality of pixels can be determined by determining a logarithmic transformation of a normalized pixel value. The normalized pixel value can correspond to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are aligned with respect to one or more control genomic regions.
[00208] In one or more examples, the group of sequence representations used to generate the one or more images can be determined by analyzing the one or more molecular characteristics with respect to one or more criteria. For example, determining the group of sequence representations can include analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region. In one or more additional examples, determining the group of sequence representations can include analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having no greater than a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region.
[00209] At 410, the process 400 can include providing the one or more images to a convolutional neural network. The convolutional neural network can computationally analyze the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects. In one or more examples, the convolutional neural network can determine multiple tumor indications that each correspond to an individual cancer type of a plurality of cancer types, in these scenarios, the multiple tumor indications can be determined by one or more output layers of the convolutional neural network. The convolutional neural network can determine a plurality of probabilities of the plurality of cancer types being present in the one more subjects. For individual output layers, the convolutional neural network can generate an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in the one or more subjects. In one or more illustrative examples, the plurality of probabilities generated by the output layers of the convolutional neural network can be analyzed to determine a type of cancer of the plurality of types of cancer having a highest probability of being present in the one or more subjects.
[00210] In various examples, image data can be generated for a plurality of genomic regions. For example, the sequencing data can be analyzed to determine a plurality of additional groups of additional sequence representations in relation to a plurality of additional genomic regions. Additionally, analyzing the plurality of additional groups of sequence representations to determine additional values of the one or more molecular characteristics for additional individual sequence representations of the plurality of additional groups of sequence representations.
[00211] A plurality of additional images based on the plurality of additional groups of sequence representations can be generated based on the molecular characteristics for the additional individual sequence representations. The plurality of additional images can include a plurality of additional pixels. Individual additional pixels of the plurality of additional pixels can comprise an additional first value that corresponds to one or more additional genomic locations and an additional second value that corresponds to the one or more molecular characteristics. The individual additional pixels can also include an additional intensity value indicating an additional number of the plurality of additional sequence representations having the additional first value and the additional second value. Each additional image of the plurality of additional images can be generated using information derived from additional sequence representations included in an additional group of sequence representations of the plurality of additional groups of sequence representations. The additional sequence representations can be aligned with an additional genomic region.
[00212] In various examples, the plurality of additional images can be provided to a plurality of additional convolutional neural networks. The plurality of additional convolutional neural networks can determine a plurality of additional tumor indications related to a tumor being present in the one or more samples. Individual additional convolutional neural networks of the plurality of additional convolutional neural networks can analyze a portion of the plurality of additional images corresponding to a given genomic region to determine an additional tumor indication related to a tumor being present in the one or more subjects. The group of tumor indications generated by the individual convolutional neural networks can be analyzed to determine an overall tumor indication related to a tumor being present in the one or more subjects. In one or more illustrative examples, the group of tumor indications can be analyzed using a logistic regression technique to determine the overall tumor indication. In at least some examples, the tumor indications determined by the convolutional neural networks corresponding to the individual genomic regions of the plurality of genomic regions can include probabilities of a tumor being present in the one or more subjects. In one or more illustrative examples, the probabilities of a tumor being present in the one or more subjects can be analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
[00213] In one or more examples, multiple images can be generated for a genomic region with each image corresponding to a different molecular characteristic. For example, the one or more images can include a first image that corresponds to the genomic region and a second image that corresponds to the genomic region. The first image can include first pixel values that comprise (i) first values that correspond to genomic locations within the genomic region and (ii) second values that correspond to a first molecular characteristic of the individual sequence representations of the group of sequence representations. Additionally, the second image can include second pixel values that comprise (i) the first values that correspond to the genomic locations within the genomic region and (ii) additional second values that correspond to a second molecular characteristic of the individual sequence representations of the group of sequence representations. The first image can be analyzed using a first convolutional neural network to determining a first tumor indication related to a tumor being present in the one or more subjects. Further, the second image can be analyzed using a second convolutional neural network to determine a second tumor indication of a tumor being present in the one or more subjects. An overall tumor indication of a tumor being present in the one or more subjects can be determined based on the first tumor indication and the second tumor indication.
[00214] The convolutional neural network can include one or more computational models that are produced as part of a training process. To illustrate, generating a trained version of the convolutional neural network can include obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected. First training images can then be produced based on the first training sequence representations. Individual first training images can include a first plurality of pixels. Individual pixels of the first plurality of pixels can comprise a first training value that corresponds to one or more genomic locations within an individual genomic region and a second training value that corresponds to the one or more molecular characteristics. The individual pixels of the first plurality of pixels can also include a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value.
[00215] The training process of the convolutional neural network can also include obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected. Second training images can be generated based on the second training sequence representations. Individual second training images can include a second plurality of pixels. In addition, individual pixels of the second plurality of pixels can comprise a first additional training value that corresponds to one or more genomic locations within an individual genomic region and a second additional training value that corresponds to the one or more molecular characteristics. Further, the individual pixel values of the second plurality of pixel values can include a second intensity training value indicating a number of the second training sequence representations having the first additional training value and the second additional training value.
[00216] In various examples, a plurality of iterations of a training process can be performed for the convolutional neural network to determine weights of layers of the convolutional neural network. Individual iterations of the plurality of iterations of the training process for the convolutional neural network can include determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network. Additionally, individual iterations of the plurality of iterations of the training process for the convolutional neural network can include determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network. In one or more examples, the training process for the convolutional neural network can include determining differences between the first weights and the second weights and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights. In at least some examples, a training process for the convolutional neural network can be performed using samples from subjects in which cancer is present and samples from subjects in which cancer is not detected, during one or more iterations of the training process, a loss is calculated for individual sets of training data that include a portion of the training data derived from samples obtained from subjects in which cancer is present and a portion of the training data derived from samples obtained from subjects in which cancer is not detected. Weights for the convolutional neural network are updated based on gradients that are determined from differences in the loss values between iterations of the training process. EXEMPLARY METHODS
A. Determining an indication of a tumor-related biological condition in a sample
[00217] The techniques described herein relate to one or more methods that include obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects. The method can also include computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations. The group of sequence representations can correspond to one or more portions of a genomic region. In addition, the method can include computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations. Further, the method can include generating, based on the group of sequence representations, one or more images that include a plurality of pixels. Individual pixels of the plurality of pixels can include (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value. The method can include providing the one or more images to a convolutional neural network. The convolutional neural network can computationally analyze the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
[00218] The techniques described herein relate to one or more computing apparatus that include: one or more hardware processors; and memory storing computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects. The memory can also store additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations. The group of sequence representations can correspond to one or more portions of a genomic region. Additionally, the memory can store additional computer- readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations and generating, based on the group of sequence representations, one or more images that include a plurality of pixels. Individual pixels of the plurality of pixels can comprise (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value. Further, the memory can store additional computer- readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising providing the one or more images to a convolutional neural network. The convolutional neural network can computationally analyze the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
[00219] The techniques described herein relate to one or more non-transitory computer-readable media storing computer-readable instructions that, when executed by one or more hardware processors, cause the one or more hardware processors to perform operations comprising obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects. The one or more non-transitory computer- readable media can also store additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations. The group of sequence representations can correspond to one or more portions of a genomic region. Additionally, the one or more non-transitory computer-readable media can also store additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations and generating, based on the group of sequence representations, one or more images that include a plurality of pixels. Individual pixels of the plurality of pixels can comprise (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value. Further, the one or more non-transitory computer-readable media can also store additional computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising providing the one or more images to a convolutional neural network. The convolutional neural network can computationally analyze the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
B. Partitioning the sample into a plurality of subsamples
[00220] In some embodiments described herein, different forms of DNA (e.g., hypermethylated and hypomethylated DNA) are physically partitioned based on one or more characteristics of the DNA. This approach can be used to determine, for example, whether certain sites or regions are hypermethylated or hypomethylated. Partitioning can be performed before attaching adapters to DNA molecules in the sample, e.g., so as to facilitate including partition tags in the adapters. Partition tags can be used to identify which partition a molecule was found in. Following partitioning (and attachment of adapters if applicable), further steps such as amplification, target capture, and sequencing may be performed. [00221] Methylation profiling can involve determining methylation patterns across different regions of the genome. For example, after partitioning molecules based on extent of methylation (e.g., relative number of methylated nucleobases per molecule) and further steps as discussed above including sequencing, the sequences of molecules in the different partitions can be mapped to a reference genome. This can show regions of the genome that, compared with other regions, are more highly methylated or are less highly methylated. In this way, genomic regions, in contrast to individual molecules, may differ in their extent of methylation.
[00222] Partitioning nucleic acid molecules in a sample can increase a rare signal, e.g., by enriching rare nucleic acid molecules that are more prevalent in one partition of the sample. For example, a genetic variation present in hypermethylated DNA but less (or not) present in hypomethylated DNA can be more easily detected by partitioning a sample into hypermethylated and hypomethylated nucleic acid molecules. By analyzing multiple partitions of a sample, a multi-dimensional analysis of a single molecule can be performed and hence, greater sensitivity can be achieved. Partitioning may include physically partitioning nucleic acid molecules into partitions or subsamples based on the presence or absence of one or more methylated nucleobases. A sample may be partitioned into partitions or subsamples based on a characteristic that is indicative of differential gene expression or a disease state. A sample may be partitioned based on a characteristic, or combination thereof that provides a difference in signal between a normal and diseased state during analysis of nucleic acids, e.g., cell free DNA (cfDNA), non-cfDNA, tumor DNA, circulating tumor DNA (ctDNA) and cell free nucleic acids (cfNA). [00223] In some embodiments, hypermethylation and/or hypomethylation variable epigenetic target regions are analyzed to determine whether they show differential methylation characteristic of particular immune cell types, such as rare immune cell types, tumor cells or cells of a type that does not normally contribute to the DNA sample being analyzed (such as cfDNA).
[00224] In some instances, heterogeneous DNA in a sample is partitioned into two or more partitions (e.g., at least 3, 4, 5, 6 or 7 partitions). In some embodiments, each partition is differentially tagged. Tagged partitions can then be pooled together for collective sample prep and/or sequencing. The partitioning-tagging-pooling steps can occur more than once, with each round of partitioning occurring based on a different characteristic (examples provided herein), and tagged using differential tags that are distinguished from other partitions and partitioning means. In other instances, the differentially tagged partitions are separately sequenced.
[00225] In some embodiments, sequence reads from differentially tagged and pooled DNA are obtained and analyzed in silico. Tags are used to sort reads from different partitions. Analysis to detect genetic variants can be performed on a partition-by-partition level, as well as whole nucleic acid population level. For example, analysis can include in silico analysis to determine genetic variants, such as CNV, SNV, indel, fusion in nucleic acids in each partition. In some instances, in silico analysis can include determining chromatin structure. For example, coverage of sequence reads can be used to determine nucleosome positioning in chromatin. Higher coverage can correlate with higher nucleosome occupancy in genomic region while lower coverage can correlate with lower nucleosome occupancy or nucleosome depleted region (NDR).
[00226] In some embodiments, partitioning is on the basis of one or more characteristics such as methylation. Molecules can be sorted according to other characteristics, such as sequence length, nucleosome binding, sequence mismatch, immunoprecipitation, and/or proteins that bind to DNA, using appropriate techniques as part of data analysis or partitioning as applicable. Resulting partitions can include one or more of the following nucleic acid forms: single-stranded DNA (ssDNA), double-stranded DNA (dsDNA), shorter DNA fragments and longer DNA fragments. In some embodiments, partitioning based on a cytosine modification (e.g., cytosine methylation) or methylation generally is performed and is optionally combined with at least one additional partitioning step, which may be based on any of the foregoing characteristics or forms of DNA. In some embodiments, a heterogeneous population of nucleic acids is partitioned into nucleic acids with one or more epigenetic modifications and without the one or more epigenetic modifications. Examples of epigenetic modifications include presence or absence of methylation; level of methylation; type of methylation (e.g., 5- methylcytosine versus other types of methylation, such as adenine methylation and/or cytosine hydroxymethylation); and association and level of association with one or more proteins, such as histones. Alternatively, or additionally, a heterogeneous population of nucleic acids can be partitioned into nucleic acid molecules associated with nucleosomes and nucleic acid molecules devoid of nucleosomes. Alternatively, or additionally, a heterogeneous population of nucleic acids may be partitioned into single-stranded DNA (ssDNA) and double-stranded DNA (dsDNA). Alternatively, or additionally, a heterogeneous population of nucleic acids may be partitioned based on nucleic acid length (e.g., molecules of up to 160 bp and molecules having a length of greater than 160 bp).
[00227] The agents used to partition populations of nucleic acids within a sample can be affinity agents, such as antibodies with the desired specificity, natural binding partners or variants thereof (Bock et al. , Nat Biotech 28: 1106-1114 (2010); Song et al. , Nat Biotech 29: 68-72 (2011 )), or artificial peptides selected e.g., by phage display to have specificity to a given target. In some embodiments, the agent used in the partitioning is an agent that recognizes a modified nucleobase. In some embodiments, the modified nucleobase recognized by the agent is a modified cytosine, such as a methylcytosine (e.g., 5-methylcytosine). In some embodiments, the modified nucleobase recognized by the agent is a product of a procedure that affects the first nucleobase in the DNA differently from the second nucleobase in the DNA of the sample. In some embodiments, the modified nucleobase may be a “converted nucleobase,” meaning that its base pairing specificity was changed by the procedure. For example, certain procedures convert unmethylated or unmodified cytosine to dihydrouracil, or more generally, at least one modified or unmodified form of cytosine undergoes deamination, resulting in uracil (considered a modified nucleobase in the context of DNA) or a further modified form of uracil. Examples of partitioning agents include antibodies, such as antibodies that recognize a modified nucleobase, which may be a modified cytosine, such as a methylcytosine (e.g., 5-methylcytosine). In some embodiments, the partitioning agent is an antibody that recognizes a modified cytosine other than 5-methylcytosine, such as 5- carboxylcytosine (5caC). Alternative partitioning agents include methyl binding domain (MBDs) and methyl binding proteins (MBPs) as described herein, including proteins such as MeCP2.
[00228] Additional, non-limiting examples of partitioning agents are histone binding proteins which can separate nucleic acids bound to histones from free or unbound nucleic acids. Examples of histone binding proteins that can be used in the methods disclosed herein include RBBP4, RbAp48 and SANT domain peptides.
[00229] The binding of partitioning agents to particular nucleic acids and the partitioning of the nucleic acids into subsamples may occur to a certain extent or may occur in an essentially binary manner. In some instances, nucleic acids comprising a greater proportion of a certain modification bind to the agent at a greater extent than nucleic acids comprising a lesser proportion of the modification. Similarly, the partitioning may produce subsamples comprising greater and lesser proportions of nucleic acids comprising a certain modification. Alternatively, the partitioning may produce subsamples comprising essentially all or none of the nucleic acids comprising the modification. In all instances, various levels of modifications may be sequentially eluted from the partitioning agent.
[00230] In some embodiments, partitioning can comprise both binary partitioning and partitioning based on degree/level of modifications. For example, methylated fragments can be partitioned by methylated DNA immunoprecipitation (MeDIP), or all methylated fragments can be partitioned from unmethylated fragments using methyl binding domain proteins (e.g., MethylMinder Methylated DNA Enrichment Kit (ThermoFisher Scientific). Subsequently, additional partitioning may involve eluting fragments having different levels of methylation by adjusting the salt concentration in a solution with the methyl binding domain and bound fragments. As salt concentration increases, fragments having greater methylation levels are eluted.
[00231] In some instances, the final partitions are enriched in nucleic acids having different extents of modifications (overrepresentative or underrepresentative of modifications). Overrepresentation and underrepresentation can be defined by the number of modifications bom by a nucleic acid relative to the median number of modifications per strand in a population. For example, if the median number of 5- methylcytosine residues in nucleic acid in a sample is 2, a nucleic acid including more than two 5-methylcytosine residues is overrepresented in this modification and a nucleic acid with 1 or zero 5-methylcytosine residues is underrepresented. The effect of the affinity separation is to enrich for nucleic acids overrepresented in a modification in a bound phase and for nucleic acids underrepresented in a modification in an unbound phase (i.e., in solution). The nucleic acids in the bound phase can be eluted before subsequent processing.
[00232] When using MeDIP or MethylMinerOMethylated DNA Enrichment Kit (ThermoFisher Scientific) various levels of methylation can be partitioned using sequential elutions. For example, a hypomethylated partition (no methylation) can be separated from a methylated partition by contacting the nucleic acid population with the MBD from the kit, which is attached to magnetic beads. The beads are used to separate out the methylated nucleic acids from the non- methylated nucleic acids. Subsequently, one or more elution steps are performed sequentially to elute nucleic acids having different levels of methylation. For example, a first set of methylated nucleic acids can be eluted at a salt concentration of 160 mM or higher, e.g., at least 150 mM, at least 200 mM, 300 mM, 400 mM, 500 mM, 600 mM, 700 mM, 800 mM, 900 mM, 1000 mM, or 2000 mM. After such methylated nucleic acids are eluted, magnetic separation is once again used to separate higher level of methylated nucleic acids from those with lower level of methylation. The elution and magnetic separation steps can be repeated to create various partitions such as a hypomethylated partition (enriched in nucleic acids comprising no methylation), a methylated partition (enriched in nucleic acids comprising low levels of methylation), and a hyper methylated partition (enriched in nucleic acids comprising high levels of methylation).
[00233] In some methods, nucleic acids bound to an agent used for affinity separation-based partitioning are subjected to a wash step. The wash step washes off nucleic acids weakly bound to the affinity agent. Such nucleic acids can be enriched in nucleic acids having the modification to an extent close to the mean or median (i.e., intermediate between nucleic acids remaining bound to the solid phase and nucleic acids not binding to the solid phase on initial contacting of the sample with the agent).
[00234] The affinity separation results in at least two, and sometimes three or more partitions of nucleic acids with different extents of a modification. While the partitions are still separate, the nucleic acids of at least one partition, and usually two or three (or more) partitions are linked to nucleic acid tags, usually provided as components of adapters, with the nucleic acids in different partitions receiving different tags that distinguish members of one partition from another. The tags linked to nucleic acid molecules of the same partition can be the same or different from one another. But if different from one another, the tags may have part of their code in common so as to identify the molecules to which they are attached as being of a particular partition.
[00235] For further details regarding portioning nucleic acid samples based on characteristics such as methylation, see WO2018/119452, which is incorporated herein by reference.
[00236] In some embodiments, the nucleic acid molecules can be fractionated into different partitions based on the nucleic acid molecules that are bound to a specific protein or a fragment thereof and those that are not bound to that specific protein or fragment thereof.
[00237] Nucleic acid molecules can be fractionated based on DNA-protein binding. Protein-DNA complexes can be fractionated based on a specific property of a protein. Examples of such properties include various epitopes, modifications (e.g., histone methylation or acetylation) or enzymatic activity. Examples of proteins which may bind to DNA and serve as a basis for fractionation may include, but are not limited to, protein A and protein G. Any suitable method can be used to fractionate the nucleic acid molecules based on protein bound regions. Examples of methods used to fractionate nucleic acid molecules based on protein bound regions include, but are not limited to, SDS-PAGE, chromatin-immuno-precipitation (ChIP), heparin chromatography, and asymmetrical field flow fractionation (AF4).
[00238] In some embodiments, the partitioning of the sample into a plurality of subsamples is performed by contacting the nucleic acids with an antibody that recognizes a modified nucleobase in the DNA, which may be is a modified cytosine or a product of the procedure that affects the first nucleobase in the DNA differently from the second nucleobase in the DNA of the sample. In some embodiments, the modified nucleobase is 5mC. In some embodiments, the modified nucleobase is 5caC. In some embodiments, the modified nucleobase is dihydrouracil (DHU). In some embodiments, the antibody that recognizes a modified nucleobase in the DNA is used to partition single-stranded DNA.
[00239] In some embodiments, the partitioning is performed by contacting the nucleic acids with a methyl binding domain (“MBD”) of a methyl binding protein (“MBP”). In some such embodiments, the nucleic acids are contacted with an entire MBP. In some embodiments, an MBD binds to 5-methylcytosine (5mC), and an MBP comprises an MBD and is referred to interchangeably herein as a methyl binding protein or a methyl binding domain protein. In some embodiments, an MBD binds to 5mC and 5hmC. In some embodiments, MBD is coupled to paramagnetic beads, such as Dynabeads® M-280 Streptavidin via a biotin linker. Partitioning into fractions with different extents of methylation can be performed by eluting fractions by increasing the NaCI concentration. [00240] In some embodiments, bound DNA is eluted by contacting the antibody or MBD with a protease, such as proteinase K. This may be performed instead of or in addition to elution steps using NaCI as discussed above.
[00241] Examples of agents that recognize a modified nucleobase contemplated herein include, but are not limited to:
(a) MeCP2 is a protein that preferentially binds to 5-methyl-cytosine over unmodified cytosine.
(b) RPL26, PRP8 and the DNA mismatch repair protein MHS6 preferentially bind to 5- hydroxymethyl-cytosine over unmodified cytosine.
(c) FOXK1 , FOXK2, FOXP1 , FOXP4 and FOXI3 preferably bind to 5-formyl cytosine over unmodified cytosine (lurlaro et al., Genome Biol. 14: R119 (2013)).
(d) Antibodies specific to one or more methylated or modified nucleobases or conversion products thereof, such as 5mC, 5caC, or DHU.
[00242] In general, elution is a function of the number of modifications, such as the number of methylated sites per molecule, with molecules having more methylation eluting under increased salt concentrations. To elute the DNA into distinct populations based on the extent of methylation, one can use a series of elution buffers of increasing NaCI concentration. Salt concentration can range from about 100 nm to about 2500 mM NaCI. In one embodiment, the process results in three (3) partitions. Molecules are contacted with a solution at a first salt concentration and comprising a molecule comprising an agent that recognizes a modified nucleobase, which molecule can be attached to a capture moiety, such as streptavidin. At the first salt concentration a population of molecules will bind to the agent and a population will remain unbound. The unbound population can be separated as a “hypomethylated” population. For example, a first partition enriched in hypomethylated form of DNA is that which remains unbound at a low salt concentration, e.g., 100 mM or 160 mM. A second partition enriched in intermediate methylated DNA is eluted using an intermediate salt concentration, e.g., between 100 mM and 2000 mM concentration. This is also separated from the sample. A third partition enriched in hypermethylated form of DNA is eluted using a high salt concentration, e.g., at least about 2000 mM.
[00243] In some embodiments, a monoclonal antibody raised against 5- methylcytidine (5mC) is used to purify methylated DNA. DNA is denatured, e.g., at 95°C in order to yield single-stranded DNA fragments. Protein G coupled to standard or magnetic beads as well as washes following incubation with the anti-5mC antibody are used to immunoprecipitate DNA bound to the antibody. Such DNA may then be eluted. Partitions may comprise unprecipitated DNA and one or more partitions eluted from the beads.
[00244] In some embodiments, sample DNA (e.g., between 5 and 200 ng) is mixed with methyl binding domain (MBD) buffer and magnetic beads conjugated with MBD proteins and incubated overnight. Methylated DNA (hypermethylated DNA) binds the MBD protein on the magnetic beads during this incubation. Non-methylated (hypomethylated DNA) or less methylated DNA (intermediately methylated) is washed away from the beads with buffers containing increasing concentrations of salt. For example, one, two, or more fractions containing non-methylated, hypomethylated, and/or intermediately methylated DNA may be obtained from such washes. Finally, a high salt buffer is used to elute the heavily methylated DNA (hypermethylated DNA) from the MBD protein. In some embodiments, these washes result in three partitions (hypomethylated partition, intermediately methylated fraction and hypermethylated partition) of DNA having increasing levels of methylation.
[00245] In some embodiments, partitioning procedures may result in imperfect sorting of DNA molecules among the subsamples. For example, a minority of the molecules in an unmethylated or hypomethylated subsample may be highly modified (e.g., hypermethylated), and/or a minority of the molecules in a hypermethylated subsample may be unmodified or mostly unmodified (e.g., unmethylated or mostly unmethylated). Such molecules are considered nonspecifically partitioned. [00246] In some embodiments, nonspecifically partitioned molecules are removed using a methylation-dependent nuclease, e.g., a methylation dependent restriction enzyme (MDRE), digesting/cleaving the DNA where the restriction enzyme (RE) recognition site contains a methylated nucleotide but not cleaving the DNA where the restriction enzyme (RE) recognition site contains an unmethylated nucleotide. In some embodiments, nonspecifically partitioned molecules are removed using a methylation sensitive nuclease, e.g., a methylation sensitive restriction enzyme (MSRE), digesting/cleaving the DNA where the restriction enzyme (RE) recognition site contains an unmethylated nucleotide but not cleaving the DNA where the restriction enzyme (RE) recognition site contains a methylated nucleotide. For example, in some embodiments, a hypomethylated subsample is contacted with a methylation-dependent nuclease, such as a methylation-dependent restriction enzyme, thereby degrading nonspecifically partitioned DNA, e.g., methylated DNA, in the subsample. Alternatively, or in addition, a hypermethylated subsample is contacted with a methylation-sensitive endonuclease, such as a methylation-sensitive restriction enzyme, thereby degrading nonspecifically partitioned DNA in the subsample.
[00247] Degradation of nonspecifically partitioned DNA in one or more partitioned subsamples may improve the performance of methods that rely on accurate partitioning of DNA on the basis of a cytosine modification. For example, such degradation may provide improved sensitivity and/or simplify downstream analyses. In some embodiments, partitioning DNA on the basis of a modification, such as methylation, then removing nonspecifically partitioned DNA using MDREs and/or MSREs as described herein provides improved efficiency and/or cost over DNA analysis methods comprising procedures that affect a first nucleobase differently from a second nucleobase, such as bisulfite sequencing or bisulfite conversion.
[00248] In some embodiments, one or more nucleases are used to degrade nonspecifically partitioned DNA molecules. In some embodiments, a subsample is contacted with a plurality of nucleases. The subsample may be contacted with the nucleases sequentially or simultaneously. Simultaneous use of nucleases may be advantageous when the nucleases are active under similar conditions (e.g., buffer composition) to avoid unnecessary sample manipulation. Contacting a subsample with more than one methylation-dependent restriction enzyme can more completely degrade nonspecifically partitioned hypermethylated DNA. Contacting a subsample with more than one methylation-sensitive restriction enzyme can more completely degrade nonspecifically partitioned hypomethylated and/or unmethylated DNA.
[00249] In some embodiments, a methylation-dependent nuclease comprises one or more of MspJI, LpnPI, FspEI, or McrBC. In some embodiments, at least two methylation-dependent nucleases are used. In some embodiments, at least three methylation-dependent nucleases are used.
[00250] In some embodiments, a methylation-sensitive nuclease comprises one or more of Aatll, Accll, Acil, Aor13HI, Aor15HI, BspT104l, BssHII, BstUI, CfrI Ol, Clal, Cpol, Eco52l, Haell, Hapll, Hhal, Hin6l, Hpall, HpyCH4IV, Mlul, Mspl, Nael, Notl, Nrul, Nsbl, PmaCI, Psp1406l, Pvul, Sacll, Sall, Smal, and SnaBI. In some embodiments, at least two methylation-sensitive nucleases are used. In some embodiments, at least three methylation-sensitive nucleases are used. In some embodiments, the methylationsensitive nucleases comprise BstUI and Hpall. In some embodiments, the two methylation-sensitive nucleases comprise Hhal and Accll. In some embodiments, the methylation-sensitive nucleases comprise BstUI, Hpall and Hin6l.
[00251] In some embodiments, the partitions of DNA are desalted and concentrated in preparation for enzymatic steps of library preparation.
C. Adapter Ligation
[00252] In some embodiments, adapters are added to the DNA. This may be done concurrently with an amplification procedure, e.g., by providing the adapters in a 5’ portion of a primer (where PCR is used, this can be referred to as library prep-PCR or LP-PCR). In some embodiments, adapters are added by other approaches, such as ligation. In some such methods, prior to partitioning or prior to capturing, first adapters are added to the nucleic acids by ligation to the 3’ ends thereof, which may include ligation to singlestranded DNA. The adapter can be used as a priming site for second-strand synthesis, e.g., using a universal primer and a DNA polymerase. A second adapter can then be ligated to at least the 3’ end of the second strand of the now double-stranded molecule. In some embodiments, the first adapter comprises an affinity tag, such as biotin, and nucleic acid ligated to the first adapter is bound to a solid support (e.g., bead), which may comprise a binding partner for the affinity tag such as streptavidin. For further discussion of a related procedure, see Gansauge et al., Nature Protocols 8:737-748 (2013). Commercial kits for sequencing library preparation compatible with single-stranded nucleic acids are available, e.g., the Accel-NGS® Methyl-Seq DNA Library Kit from Swift Biosciences. In some embodiments, after adapter ligation, nucleic acids are amplified.
[00253] Preferably, the adapters include different tags of sufficient numbers that the number of combinations of tags results in a low probability e.g., 95, 99 or 99.9% of two nucleic acids with the same start and stop points receiving the same combination of tags. Adapters, whether bearing the same or different tags, can include the same or different primer binding sites, but preferably adapters include the same primer binding site.
[00254] In some embodiments, following attachment of adapters, the nucleic acids are subject to amplification. The amplification can use, e.g., universal primers that recognize primer binding sites in the adapters.
[00255] In some embodiments, following attachment of adapters, the DNA is partitioned, comprising contacting the DNA with an agent that preferentially binds to nucleic acids bearing an epigenetic modification. The nucleic acids are partitioned into at least two subsamples differing in the extent to which the nucleic acids bear the modification from binding to the agents. For example, if the agent has affinity for nucleic acids bearing the modification, nucleic acids overrepresented in the modification (compared with median representation in the population) preferentially bind to the agent, whereas nucleic acids underrepresented for the modification do not bind or are more easily eluted from the agent. The nucleic acids can then be amplified from primers binding to the primer binding sites within the adapters. Partitioning may be performed instead before adapter attachment, in which case the adapters may comprise differential tags that include a component that identifies which partition a molecule occurred in. [0214] In some embodiments, the nucleic acids are linked at both ends to Y-shaped adapters including primer binding sites and tags. The molecules are amplified.
D. Tagging
[00256] “Tagging” DNA molecules is a procedure in which a tag is attached to or associated with the DNA molecules. Tags can be molecules, such as nucleic acids, containing information that indicates a feature of the molecule with which the tag is associated. For example, molecules can bear a sample tag (which distinguishes molecules in one sample from those in a different sample) or a molecular tag/molecular barcode/barcode (which distinguishes different molecules from one another (in both unique and non-unique tagging scenarios). For methods that involve a partitioning step, a partition tag (which distinguishes molecules in one partition from those in a different partition) may be included. In some embodiments, adapters added to DNA molecules comprise tags. In certain embodiments, a tag can comprise one or a combination of barcodes. As used herein, the term “barcode” refers to a nucleic acid molecule having a particular nucleotide sequence, or to the nucleotide sequence, itself, depending on context. A barcode can have, for example, between 10 and 100 nucleotides. A collection of barcodes can have degenerate sequences or can have sequences having a certain hamming distance, as desired for the specific purpose. So, for example, a molecular barcode can be comprised of one barcode or a combination of two barcodes, each attached to different ends of a molecule. Additionally, or alternatively, for different partitions and/or samples, different sets of molecular barcodes, or molecular tags can be used such that the barcodes serve as a molecular tag through their individual sequences and also serve to identify the partition and/or sample to which they correspond based the set of which they are a member.
[00257] In some embodiments, two or more partitions, e.g., each partition, is/are differentially tagged. Tags can be used to label the individual polynucleotide population partitions so as to correlate the tag (or tags) with a specific partition. Alternatively, tags can be used in embodiments that do not employ a partitioning step. In some embodiments, a single tag can be used to label a specific partition. In some embodiments, multiple different tags can be used to label a specific partition. In embodiments employing multiple different tags to label a specific partition, the set of tags used to label one partition can be readily differentiated for the set of tags used to label other partitions. In some embodiments, the tags may have additional functions, for example the tags can be used to index sample sources or used as unique molecular identifiers (which can be used to improve the quality of sequencing data by differentiating sequencing errors from mutations, for example as in Kinde et al. , Proc Nat’l Acad Sci USA 108: 9530-9535 (2011 ), Kou et al., PLoS ONE, 11 : eO146638 (2016)) or used as non-unique molecule identifiers, for example as described in US Pat. No. 9,598,731 . Similarly, in some embodiments, the tags may have additional functions, for example the tags can be used to index sample sources or used as non-unique molecular identifiers (which can be used to improve the quality of sequencing data by differentiating sequencing errors from mutations).
[00258] In some embodiments, partition tagging comprises tagging molecules in each partition with a partition tag. After re-combining partitions (e.g., to reduce the number of sequencing runs needed and avoid unnecessary cost) and sequencing molecules, the partition tags identify the source partition. In some embodiments, the partition tags can serve as identifiers of the source partition and the molecule, i.e. , different partitions are tagged with different sets of molecular tags, e.g., comprised of a pair of barcodes. In this way, the one or more molecular barcodes attached to the molecule indicates the source partition as well as being useful to distinguish molecules within a partition. For example, a first set of 35 barcodes can be used to tag molecules in a first partition, while a second set of 35 barcodes can be used tag molecules in a second partition.
[00259] In some embodiments, after partitioning and tagging with partition tags, the molecules may be pooled for sequencing in a single run. In some embodiments, a sample tag is added to the molecules, e.g., in a step subsequent to addition of partition tags and pooling. Sample tags can facilitate pooling material generated from multiple samples for sequencing in a single sequencing run.
[00260] Alternatively, in some embodiments, partition tags may be correlated to the sample as well as the partition. As a simple example, a first tag can indicate a first partition of a first sample; a second tag can indicate a second partition of the first sample; a third tag can indicate a first partition of a second sample; and a fourth tag can indicate a second partition of the second sample.
[00261] While tags may be attached to molecules already partitioned based on one or more characteristics, the final tagged molecules in the library may no longer possess that characteristic. For example, while single stranded DNA molecules may be partitioned and tagged, the final tagged molecules in the library are likely to be double stranded. Similarly, while DNA may be subject to partition based on different levels of methylation, in the final library, tagged molecules derived from these molecules are likely to be unmethylated. Accordingly, the tag attached to molecule in the library typically indicates the characteristic of the “parent molecule” from which the ultimate tagged molecule is derived, not necessarily to characteristic of the tagged molecule, itself.
[00262] As an example, barcodes 1 , 2, 3, 4, etc. are used to tag and label molecules in the first partition; barcodes A, B, C, D, etc. are used to tag and label molecules in the second partition; and barcodes a, b, c, d, etc. are used to tag and label molecules in the third partition. Differentially tagged partitions can be pooled prior to sequencing. Differentially tagged partitions can be separately sequenced or sequenced together concurrently, e.g., in the same flow cell of an Illumina sequencer.
[00263] After sequencing, analysis of reads can be performed on a partition-by- partition level, as well as a whole DNA population level. Tags are used to sort reads from different partitions. Analysis can include in silico analysis to determine genetic and epigenetic variation (one or more of methylation, chromatin structure, etc.) using sequence information, genomic coordinates length, coverage, and/or copy number. In some embodiments, higher coverage can correlate with higher nucleosome occupancy in genomic region while lower coverage can correlate with lower nucleosome occupancy or a nucleosome depleted region (NDR).
E. Enriching/Capturing step; Amplification
[00264] Methods disclosed herein can comprise capturing DNA, such as cfDNA target regions. In some embodiments, the capturing comprises contacting the DNA with probes (e.g., oligonucleotides) specific for the target regions. Enrichment or capture may be performed on any sample or subsample described herein using any suitable approach known in the art.
[00265] In some embodiments, enrichment or capture is performed after attachment of adapters to sample molecules. In some embodiments, enrichment or capture is performed after a partitioning step. In some embodiments, enrichment or capture is performed after an amplification step. In some embodiments, sample molecules are partitioned, then adapters are attached, then sample molecules are amplified, and then the amplified molecules are subjected to enrichment or capture. The enriched or captured molecules may then be subjected to another amplification and then sequenced. [00266] In some embodiments, the probes specific for the target regions comprise a capture moiety that facilitates the enrichment or capture of the DNA hybridized to the probes. In some embodiments, the capture moiety is biotin. In some such embodiments, streptavidin attached to a solid support, such as magnetic beads, is used to bind to the biotin. Nonspecifically bound DNA that does not comprise a target region is washed away from the captured DNA. In some embodiments, DNA is then dissociated from the probes and eluted from the solid support using salt washes or buffers comprising another DNA denaturing agent. In some embodiments, the probes are also eluted from the solid support by, e.g., disrupting the biotin-streptavidin interaction. In some embodiments, captured DNA is amplified following elution from the solid support. In some such embodiments, DNA comprising adapters is amplified using PCR primers that anneal to the adapters. In some embodiments, captured DNA is amplified while attached to the solid support. In some such embodiments, the amplification comprises use of a PCR primer that anneals to a sequence within an adapter and a PCR primer that anneals to a sequence within a probe annealed to the target region of the DNA.
[00267] In some embodiments, the methods herein comprise enriching for or capturing DNA comprising epigenetic and/or sequence-variable target regions. Such regions may be captured from an aliquot of a sample (e.g., a sample that has undergone attachment of adapters and amplification), while the step of partitioning the DNA with an agent that recognizes a modified cytosine, such as methyl cytosine, is performed on a separate aliquot of the sample. Enriching for or capturing DNA comprising epigenetic and/or sequence-variable target regions may comprise contacting the DNA with a first or second set of target-specific probes. Such target-specific probes may have any of the features described herein for sets of target-specific probes, including but not limited to in the embodiments set forth above and the sections relating to probes below. Capturing may be performed on one or more subsamples prepared during methods disclosed herein. In some embodiments, DNA is captured from the first subsample or the second subsample, e.g., the first subsample and the second subsample. In some embodiments, the subsamples are differentially tagged (e.g., as described herein) and then pooled before undergoing capture. Exemplary methods for capturing DNA comprising epigenetic and/or sequence-variable target regions can be found in, e.g., WO 2020/160414, which is hereby incorporated by reference.
[00268] The capturing step may be performed using conditions suitable for specific nucleic acid hybridization, which generally depend to some extent on features of the probes such as length, base composition, etc. Those skilled in the art will be familiar with appropriate conditions given general knowledge in the art regarding nucleic acid hybridization. In some embodiments, complexes of target-specific probes and DNA are formed.
[00269] In some embodiments, methods described herein comprise capturing a plurality of sets of target regions of cfDNA obtained from a subject. The target regions may comprise differences depending on whether they originated from a tumor or from healthy cells or from a certain cell type. The capturing step produces a captured set of cfDNA molecules. In some embodiments, cfDNA molecules corresponding to a sequence-variable target region set are captured at a greater capture yield in the captured set of cfDNA molecules than cfDNA molecules corresponding to an epigenetic target region set. In some embodiments, a method described herein comprises contacting cfDNA obtained from a subject with a set of target-specific probes, wherein the set of target-specific probes is configured to capture cfDNA corresponding to the sequencevariable target region set at a greater capture yield than cfDNA corresponding to the epigenetic target region set. For additional discussion of capturing steps, capture yields, and related aspects, see W02020/160414, which is incorporated herein by reference for all purposes.
[00270] It can be beneficial to capture cfDNA corresponding to the sequencevariable target region set at a greater capture yield than cfDNA corresponding to the epigenetic target region set because a greater depth of sequencing may be necessary to analyze the sequence-variable target regions with sufficient confidence or accuracy than may be necessary to analyze the epigenetic target regions. The volume of data needed to determine fragmentation patterns (e.g., to test for perturbation of transcription start sites or CTCF binding sites) or fragment abundance (e.g., in hypermethylated and hypomethylated partitions) is generally less than the volume of data needed to determine the presence or absence of cancer-related sequence mutations. Capturing the target region sets at different yields can facilitate sequencing the target regions to different depths of sequencing in the same sequencing run (e.g., using a pooled mixture and/or in the same sequencing cell).
[00271] In some embodiments, the DNA is amplified. In some embodiments, amplification is performed before the capturing step. In some embodiments, amplification is performed after the capturing step. In some embodiments, amplification is performed before and after the capturing step. In various embodiments, the methods further comprise sequencing the captured DNA, e.g., to different degrees of sequencing depth for the epigenetic and sequence-variable target region sets, consistent with the discussion herein.
[00272] In some embodiments, a capturing step is performed with probes for a sequence-variable target region set and probes for an epigenetic target region set in the same vessel at the same time, e.g., the probes for the sequence-variable and epigenetic target region sets are in the same composition. This approach provides a relatively streamlined workflow. In some embodiments, the concentration of the probes for the sequence-variable target region set is greater that the concentration of the probes for the epigenetic target region set.
[00273] Alternatively, a capturing step is performed with a sequence-variable target region probe set in a first vessel and with an epigenetic target region probe set in a second vessel, or a contacting step is performed with a sequence-variable target region probe set at a first time and a first vessel and an epigenetic target region probe set at a second time before or after the first time. This approach allows for preparation of separate first and second compositions comprising captured DNA corresponding to a sequencevariable target region set and captured DNA corresponding to an epigenetic target region set. The compositions can be processed separately as desired (e.g., to partition based on methylation as described herein) and pooled in appropriate proportions to provide material for further processing and analysis such as sequencing.
[00274] In some embodiments, adapters are included in the DNA as described herein. In some embodiments, tags, which may be or include barcodes, are included in the DNA. In some embodiments, such tags are included in adapters. Tags can facilitate identification of the origin of a nucleic acid. For example, barcodes can be used to allow the origin (e.g., subject) whence the DNA came to be identified following pooling of a plurality of samples for parallel sequencing. This may be done concurrently with an amplification procedure, e.g., by providing the barcodes in a 5’ portion of a primer, e.g., as described herein. In some embodiments, adapters and tags/barcodes are provided by the same primer or primer set. For example, the barcode may be located 3’ of the adapter and 5’ of the target-hybridizing portion of the primer. Alternatively, barcodes can be added by other approaches, such as ligation, optionally together with adapters in the same ligation substrate.
[00275] Additional details regarding amplification, tags, and barcodes are discussed herein, which can be combined to the extent practicable with any of these embodiments.
F. Procedures that affect a first nucleobase in the DNA differently from a second nucleobase in the DNA or methylation-sensitive conversion methods
[00276] In some embodiments, methods disclosed herein comprise a step of subjecting DNA, or a subsample thereof, to a procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA, wherein the first nucleobase is a modified or unmodified nucleobase, the second nucleobase is a modified or unmodified nucleobase different from the first nucleobase, and the first nucleobase and the second nucleobase have the same base pairing specificity. In some embodiments, the procedure chemically converts the first or second nucleobase such that the base pairing specificity of the converted nucleobase is altered. In some embodiments, DNA is subjected to a procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA before library preparation using the DNA, before a first amplification of the DNA, before dividing the DNA into a plurality of subsamples, or any combination thereof. In certain embodiments, the DNA is subjected to the procedure before or after contacting the DNA with a methylation-sensitive nuclease.
[00277] In some embodiments, the procedure that affects a first nucleobase of the DNA differently from a second nucleobase of the DNA is performed prior to the sequencing and/or (a) prior to or after the selectively depleting the target nucleic acid comprising the wild-type sequence, the target nucleic acid comprising the converted nucleotide, or the target nucleic acid that does not comprise the converted nucleotide; (b) prior to the amplifying the selectively digested population of target nucleic acids; (c) prior to or after the partitioning the population of target nucleic acids into a plurality of subsamples; and/or (d) prior to or after a step of enriching for one or more sets of target regions of DNA.
[00278] In some embodiments, if the first nucleobase is a modified or unmodified adenine, then the second nucleobase is a modified or unmodified adenine; if the first nucleobase is a modified or unmodified cytosine, then the second nucleobase is a modified or unmodified cytosine; if the first nucleobase is a modified or unmodified guanine, then the second nucleobase is a modified or unmodified guanine; and if the first nucleobase is a modified or unmodified thymine, then the second nucleobase is a modified or unmodified thymine (where modified and unmodified uracil are encompassed within modified thymine for the purpose of this step).
[00279] In some embodiments, the first nucleobase is a modified or unmodified cytosine, then the second nucleobase is a modified or unmodified cytosine. For example, first nucleobase may comprise unmodified cytosine (C) and the second nucleobase may comprise one or more of 5-methylcytosine (mC) and 5-hydroxymethylcytosine (hmC). Alternatively, the second nucleobase may comprise C and the first nucleobase may comprise one or more of mC and hmC. Other combinations are also possible, such as where one of the first and second nucleobases comprises mC and the other comprises hmC.
[00280] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises bisulfite conversion. Treatment with bisulfite converts unmodified cytosine and certain modified cytosine nucleotides (e.g. 5-formyl cytosine (fC) or 5-carboxylcytosine (caC)) to uracil whereas other modified cytosines (e.g., 5-methylcytosine, 5-hydroxylmethylcystosine) are not converted. Thus, where bisulfite conversion is used, the first nucleobase comprises one or more of unmodified cytosine, 5-formyl cytosine, 5-carboxylcytosine, or other cytosine forms affected by bisulfite, and the second nucleobase may comprise one or more of mC and hmC, such as mC and optionally hmC. Sequencing of bisulfite-treated DNA identifies positions that are read as cytosine as being mC or hmC positions. Meanwhile, positions that are read as T are identified as being T or a bisulfite-susceptible form of C, such as unmodified cytosine, 5-formyl cytosine, or 5-carboxylcytosine. Performing bisulfite conversion, such as on a DNA sample as described herein, facilitates identifying positions containing mC or hmC using the sequence reads obtained from the exemplary sample. For an exemplary description of bisulfite conversion, see, e.g., Moss et al., Nat Commun. 2018; 9: 5068.
[00281] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises oxidative bisulfite (Ox- BS) conversion. This procedure first converts hmC to fC, which is bisulfite susceptible, followed by bisulfite conversion. Thus, when oxidative bisulfite conversion is used, the first nucleobase comprises one or more of unmodified cytosine, fC, caC, hmC, or other cytosine forms affected by bisulfite, and the second nucleobase comprises mC. Sequencing of Ox-BS converted DNA identifies positions that are read as cytosine as being mC positions. Meanwhile, positions that are read as T are identified as being T, hmC, or a bisulfite-susceptible form of C, such as unmodified cytosine, fC, or hmC. Performing Ox-BS conversion, such as on a DNA sample as described herein, thus facilitates identifying positions containing mC using the sequence reads obtained from the sample. For an exemplary description of oxidative bisulfite conversion, see, e.g., Booth et al., Science 2012; 336: 934-937.
[00282] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises Tet-assisted bisulfite (TAB) conversion. In TAB conversion, hmC is protected from conversion and mC is oxidized in advance of bisulfite treatment, so that positions originally occupied by mC are converted to U while positions originally occupied by hmC remain as a protected form of cytosine. For example, as described in Yu et al., Cell 2012; 149: 1368-80, [3-glucosyl transferase can be used to protect hmC (forming 5-glucosylhydroxymethylcytosine (ghmC)), then a TET protein such as mTetl can be used to convert mC to caC, and then bisulfite treatment can be used to convert C and caC to II while ghmC remains unaffected. [00283] Alternatively, a carbamoyltransferase enzyme, such as 5- hydroxymethylcytosine carbamoyltransferase as described in Yang et al., Bio-protocol, 2023; 12(17): e4496, can be used to protect hmC (by converting hmC to 5- carbamoyloxymethylcytosine (5cmC)), then a TET protein such as mTetl or a TET2 comprising a T1372S mutation, can be used to convert mC to caC, and then bisulfite treatment can be used to convert C and caC to II while 5cmC remains unaffected. Thus, when TAB conversion is used, the first nucleobase comprises one or more of unmodified cytosine, fC, caC, mC, or other cytosine forms affected by bisulfite, and the second nucleobase comprises hmC. Sequencing of TAB-converted DNA identifies positions that are read as cytosine as being hmC positions. Meanwhile, positions that are read as T are identified as being T, mC, or a bisulfite-susceptible form of C, such as unmodified cytosine, fC, or caC. Performing TAB conversion, such as on a DNA sample as described herein, thus facilitates identifying positions containing hmC using the sequence reads obtained from the sample.
[00284] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises Tet-assisted conversion with a substituted borane reducing agent, optionally wherein the substituted borane reducing agent is 2-picoline borane, borane pyridine, tert-butylamine borane, or ammonia borane. In Tet-assisted pic-borane conversion with a substituted borane reducing agent conversion, a TET protein is used to convert mC and hmC to caC, without affecting unmodified C. caC, and fC if present, are then converted to dihydrouracil (DHU) by treatment with 2-picoline borane (pic-borane) or another substituted borane reducing agent such as borane pyridine, tert-butylamine borane, or ammonia borane, also without affecting unmodified C. See, e.g., Liu et al., Nature Biotechnology 2019; 37:424-429 (e.g., at Supplementary Fig. 1 and Supplementary Note 7). Thus, when this type of conversion is used, the first nucleobase comprises one or more of 5mC, 5fC, 5caC, or 5hmC, and the second nucleobase comprises unmodified cytosine. DHU is read as a T in sequencing. Thus, when this type of conversion is used, the first nucleobase comprises one or more of mC, fC, caC, or hmC, and the second nucleobase comprises unmodified cytosine. Sequencing of the converted DNA identifies positions that are read as cytosine as being unmodified C positions. Meanwhile, positions that are read as T are identified as being T, mC, fC, caC, or hmC. Performing TAP conversion, such as on a DNA sample as described herein, thus facilitates identifying positions containing unmodified C using the sequence reads obtained from the sample. This procedure encompasses Tet- assisted pyridine borane sequencing (TAPS), described in further detail in Liu et al. 2019, supra.
[00285] Alternatively, protection of hmC (e.g., using |3GT or 5- hydroxymethylcytosine carbamoyltransferase) can be combined with Tet-assisted conversion with a substituted borane reducing agent, e.g. as described above. In this method (TAPS-|3), 5hmC can be protected from conversion, for example through glucosylation using [3-glucosyl transferase (PGT), forming 5- glucosylhydroxymethylcytosine (5ghmC), or through carbamoylation using 5- hydroxymethylcytosine carbamoyltransferase, forming 5cmC. This is described in Yu et al., Cell 2012; 149: 1368-80. Treatment with a TET protein, such as mTetl or a TET2 comprising a T1372S mutation, then converts mC to caC but does not convert C, 5ghmC, or 5cmC. 5caC is then converted to DHU by treatment with pic-borane or another substituted borane reducing agent such as borane pyridine, tert-butylamine borane, or ammonia borane, also without affecting ghmC, 5cmC, or unmodified C. Thus, when Tet- assisted conversion with a substituted borane reducing agent is used, the first nucleobase comprises mC, and the second nucleobase comprises one or more of unmodified cytosine or hmC, such as unmodified cytosine and optionally hmC, fC, and/or caC. Sequencing of the converted DNA identifies positions that are read as cytosine as being either hmC or unmodified C positions. Meanwhile, positions that are read as T are identified as being T, fC, caC, or mC. Performing TAPS[3 conversion, such as on a DNA sample as described herein, thus facilitates distinguishing positions containing unmodified C or hmC on the one hand from positions containing mC using the sequence reads obtained from the sample. For an exemplary description of this type of conversion, see, e.g., Liu et al., Nature Biotechnology 2019; 37:424-429. 5-hydroxymethylcytosine carbamoyltransferase is described in Yang et al., Bio-protocol, 2023; 12(17): e4496.
[00286] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises chemical-assisted conversion with a substituted borane reducing agent, optionally wherein the substituted borane reducing agent is 2-picoline borane, borane pyridine, tert-butylamine borane, or ammonia borane. In chemical-assisted conversion with a substituted borane reducing agent, an oxidizing agent such as potassium perruthenate (KRuO4) (also suitable for use in ox-BS conversion) is used to specifically oxidize hmC to fC. Treatment with pic-borane or another substituted borane reducing agent such as borane pyridine, tert-butylamine borane, or ammonia borane converts fC and caC to DHLI but does not affect mC or unmodified C. Thus, when this type of conversion is used, the first nucleobase comprises one or more of hmC, fC, and caC, and the second nucleobase comprises one or more of unmodified cytosine or mC, such as unmodified cytosine and optionally mC. Sequencing of the converted DNA identifies positions that are read as cytosine as being either mC or unmodified C positions. Meanwhile, positions that are read as T are identified as being T, fC, caC, or hmC. Performing this type of conversion, such as on a DNA sample as described herein, thus facilitates distinguishing positions containing unmodified C or mC on the one hand from positions containing hmC using the sequence reads obtained from the sample. For an exemplary description of this type of conversion, see, e.g., Liu et al., Nature Biotechnology 2019; 37:424-429.
[00287] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises APOBEC-coupled epigenetic (ACE) conversion. In ACE conversion, an AID/APOBEC family DNA deaminase enzyme such as APOBEC3A (A3A) is used to deaminate unmodified cytosine and mC without deaminating hmC, fC, or caC. Thus, when ACE conversion is used, the first nucleobase comprises unmodified C and/or mC (e.g., unmodified C and optionally mC), and the second nucleobase comprises hmC. Sequencing of ACE-converted DNA identifies positions that are read as cytosine as being hmC, fC, or caC positions. Meanwhile, positions that are read as T are identified as being T, unmodified C, or mC. Performing ACE conversion on a DNA sample as described herein thus facilitates distinguishing positions containing hmC from positions containing mC or unmodified C using the sequence reads obtained from the sample. For an exemplary description of ACE conversion, see, e.g., Schutsky et al., Nature Biotechnology 2018; 36: 1083-1090.
[00288] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises enzymatic conversion of the first nucleobase, e.g., as in EM-Seq. See, e.g., Vaisvila R, et al. (2019) EM-seq: Detection of DNA methylation at single base resolution from picograms of DNA. bioRxiv; DOI: 10.1101/2019.12.20.884692, available at www. biorxiv.org/content/10.1101/2019.12.20.884692v1 . For example, TET2 and T4-[3GT or 5-hydroxymethylcytosine carbamoyltransferase (described in Yang et al., Bio-protocol, 2023; 12(17): e4496) can be used to convert 5mC and 5hmC into substrates that cannot be deaminated by a deaminase (e.g., APOBEC3A), and then a deaminase (e.g., APOBEC3A) can be used to deaminate unmodified cytosines converting them to uracils. [00289] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises enzymatic conversion of the first nucleobase using a non-specific, modification-sensitive double-stranded DNA deaminase, e.g., as in SEM-seq. See, e.g., Vaisvila et al. (2023) Discovery of novel DNA cytosine deaminase activities enables a nondestructive single-enzyme methylation sequencing method for base resolution high-coverage methylome mapping of cell-free and ultra-low input DNA. bioRxiv; DOI: 10.1101/2023.06.29.547047, available at https://www.biorxiv.org/content/10.1101/2023.06.29.547047v1. SEM-Seq employs a non-specific, modification-sensitive double-stranded DNA deaminase (MsddA) in a nondestructive single-enzyme 5-methylctyosine sequencing (SEM-seq) method that deaminates unmodified cytosines. Accordingly, SEM-seq does not require the TET2 and T4-[3GT or 5-hydroxymethylcytosine carbamoyltransferase protection and denaturing steps that are of use, e.g., in APOEC3A-based protocols. Additionally, MsddA does not deaminate 5-formylated cytosines (5fC) or 5-carboxylated cytosines (5caC). In SEM-seq, unmodified cytosines in the DNA are deaminated to uracil and is read as “T” during sequencing. Modified cytosines (e.g., 5mC) are not converted and are read as “C” during sequencing. Cytosines that are read as thymines are identified as unmodified (e.g., unmethylated) cytosines or as thymines in the DNA. Performing SEM-seq conversion thus facilitates identifying positions containing 5mC using the sequence reads obtained. In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises enzymatic conversion of the first nucleobase using MsddA.
[00290] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA of the first subsample converts a modified nucleoside. In some embodiments, the conversion procedure which converts a modified nucleosides comprises enzymatic conversion, such as DM-seq, for example, as described in WO2023/288222A1 . In DM-seq, unmodified cytosines in the DNA are enzymatically protected from a subsequent deamination step wherein 5mC in 5mCpG is converted to T. The enzymatically protected unmodified (e.g., unmethylated) cytosines are not converted and are read as “C” during sequencing. Cytosines that are read as thymines (in a CpG context) are identified as methylated cytosines in the DNA. Thus, when this type of conversion is used, the first nucleobase comprises unmodified (such as unmethylated) cytosine, and the second nucleobase comprises modified (such as methylated) cytosine. Sequencing of the converted DNA identifies positions that are read as cytosine as being unmodified C positions. Meanwhile, positions that are read as T are identified as being T or 5mC. Performing DM-seq conversion thus facilitates identifying positions containing 5mC using the sequence reads obtained.
[00291] Exemplary cytosine deaminases for use herein include APOBEC enzymes, for example, APOBEC3A. Generally, AID/APOBEC family DNA deaminase enzymes such as APOBEC3A (A3A) are used to deaminate (unprotected) unmodified cytosine and 5mC. For an exemplary description of APOBEC conversion, see, e.g., Schutsky et al., Nature Biotechnology 2018; 36: 1083-1090. In some embodiments, the deaminase comprises any one or more of the deaminases or a truncated version thereof, disclosed in WO 2024/073043, which is incorporated by reference herein in its entirety.
[00292] The enzymatic protection of unmodified cytosines in the DNA comprises addition of a protective group to the unmodified cytosines. Such protective groups can comprise an alkyl group, an alkyne group, a carboxyl group, a carboxyalkyl group, an amino group, a hydroxymethyl group, a glucosyl group, a glucosylhydroxymethyl group, an isopropyl group, or a dye. For example, DNA can be treated with a methyltransferase, such as a CpG-specific methyltransferase, which adds the protective group to unmodified cytosines. The term methyltransferase is used broadly herein to refer to enzymes capable of transferring a methyl or substituted methyl (e.g., carboxymethyl) to a substrate (e.g., a cytosine in a nucleic acid). In some embodiments, the DNA is contacted with a CpG- specific DNA methyltransferase (MTase), such as a CpG-specific carboxym ethyltransferase (CxMTase), and a substituted methyl donor, such as a carboxymethyl donor (e.g., carboxymethyl-S-adenosyl-L-methionine). See, e.g., WO2021/236778A2. In particular embodiments, the CxMTase can facilitate the addition of a protective carboxymethyl group to an unmethylated cytosine. In some embodiments, the unmethylated cytosine is unmodified cytosine. The carboxymethyl group can prevent deamination of the cytosine during a deamination step (such as a deamination step using an APOBEC enzyme, such as A3A). Substituted methyl or carboxymethyl donors useful in the disclosed methods include but are not limited to, S-adenosyl-L-methionine (SAM) analogs, optionally wherein the SAM analog is carboxy-S-adenosyl-L-methionine (CxSAM). SAM analogs are described, for example, in WO2022/197593A1 . The MTase may be, for example, a CpG methyltransferase from Spiroplasma sp. strain MQ1 (M.Sssl), DNA-methyltransferase 1 (DNMT1 ), DNA-methyltransferase 3 alpha (DNMT3A), DNA-methyltransferase 3 beta (DNMT3B), or DNA adenine methyltransferase (Dam). The CxMTase may be a CpG methyltransferase from Mycoplasma penetrans (M.Mpel). In a particular embodiment, the methyltransferase enzyme is a variant of M.Mpel, wherein the amino acid corresponding to position 374 is R or K.
[00293] In one embodiment, the methyltransferase enzyme is a variant of M.Mpel having an N374R substitution or an N374K substitution. The methyltransferase variant can further comprise one or more amino acid substitutions selected from a) substitution of one or both residues T300 and E305 with S, A, G, Q, D, or N; b) substitution of one or more residues A323, N306, and Y299 with a positively charged amino acid selected from K, R or H; and/or c) substitution of S323 with A, G, K, R or H, which may enhance the activity of the enzyme.
[00294] Optionally, the conversion procedure further includes enzymatic protection of 5hmCs, such as by glucosylation of the 5hmCs (e.g., using [3GT) or by carbamoylation of the 5hmCs (e.g., using 5-hydroxymethylcytosine carbamoyltransferase), in the DNA prior to the deamination of unprotected modified cytosines. In this method, 5hmC can be protected from conversion, for example through glucosylation using [3-glucosyl transferase (PGT), forming (5-glucosylhydroxymethylcytosine) 5ghmC, or through carbamoylation using 5-hydroxymethylcytosine carbamoyltransferase, forming 5cmC. This is described, for example, in Yu et al., Cell 2012; 149: 1368-80, and in Yang et al., Bio-protocol, 2023; 12(17): e4496. Glucosylation or carbamoylation of 5hmC can reduce or eliminate deamination of 5hmC by a deaminase such as APOBEC3A. Treatment with an MTase or CxMTase then adds a protecting group to unmodified (unmethylated) cytosines in the DNA. 5mC (but not protected, unmodified cytosine and not 5ghmC or 5cmC) is then deaminated (converted to T in the case of 5mC) by treatment with a deaminase, for example, an APOBEC enzyme (such as APOBEC3A). Sequencing of the converted DNA identifies positions that are read as cytosine as being either 5hmC or unmodified C positions. Meanwhile, positions that are read as T are identified as being T or 5mC. Performing DM-seq conversion with glucosylation of 5hmC on a sample as described herein thus facilitates distinguishing positions containing unmodified C or 5hmC on the one hand from positions containing 5mC using the sequence reads obtained.
[00295] Also provided herein are methods in which alternative base conversion schemes are used. For example, unmethylated cytosines can be left intact while methylated cytosines and hydroxymethylcytosines are converted to a base read as a thymine (e.g., uracil, thymine, or dihydrouracil).
[00296] In some embodiments, methylating a cytosine in at least one first complementary strand or second complementary strand comprises contacting the cytosine with a methyltransferase such as DNMT1 or DNMT5. In such embodiments, the step of oxidizing a 5-hydroxymethylated cytosine to 5-formylcytosine (such as by contacting the 5-hydroxymethyl cytosine in a first strand and a second strand with KRuO4) can be optional.
[00297] In some embodiments, converting the modified cytosine in at least one first or second strand to a thymine or a base read as thymine comprises oxidizing a hydroxymethyl cytosine, e.g., the hydroxymethyl cytosine is oxidized to formylcytosine. In some embodiments, oxidizing the hydroxymethyl cytosine to formylcytosine comprises contacting the hydroxymethyl cytosine with a ruthenate, such as potassium ruthenate (KRuO4).
[00298] In some embodiments, the modified cytosine is converted to thymine, uracil, or dihydrouracil. In any such embodiments, amplification methods may comprise uracil- and/or dihydrouracil-tolerant amplification methods, such as PCR using a uracil- and/or dihydrouracil-tolerant DNA polymerase. [00299] In some embodiments, the method comprises converting a formylcytosine and/or a methylcytosine to carboxylcytosine as part of converting the modified cytosine in at least one first or second strand to a thymine or a base read as thymine. For example, converting the formylcytosine and/or the methylcytosine to carboxylcytosine can comprise contacting the formylcytosine and/or the methylcytosine with a TET enzyme, such as TET1 , TET2, TET3, or a TET2 comprising a T1372S mutation. In some embodiments, the method comprises reducing the carboxylcytosine as part of converting the modified cytosine in at least one first or second strand to a thymine or a base read as thymine, and/or the carboxylcytosine is reduced to dihydrouracil. In some embodiments, reducing the carboxylcytosine comprises contacting the carboxylcytosine with a borane or borohydride reducing agent.
[00300] In some embodiments, the borane or borohydride reducing agent comprises pyridine borane, 2-picoline borane, borane, tert-butylamine borane, ammonia borane, sodium borohydride, sodium cyanoborohydride (NaBH3CN), lithium borohydride (LiBH4), ethylenediamine borane, dimethylamine borane, sodium triacetoxyborohydride, morpholine borane, 4-methylmorpholine borane, trimethylamine borane, dicyclohexylamine borane, or a salt thereof. In other embodiments, the reducing agent comprises lithium aluminum hydride, sodium amalgam, amalgam, sulfur dioxide, dithionate, thiosulfate, iodide, hydrogen peroxide, hydrazine, diisobutylaluminum hydride, oxalic acid, carbon monoxide, cyanide, ascorbic acid, formic acid, dithiothreitol, betamercaptoethanol, or any combination thereof.
[00301] Various TET enzymes may be used in the disclosed methods as appropriate. In some embodiments, the one or more TET enzymes comprise TETv. TETv is described in US Patent 10,260,088. In some embodiments, the one or more TET enzymes comprise TETcd. TETcd is described in US Patent 10,260,088. In some embodiments, the one or more TET enzymes comprise TET 1 . In some embodiments, the one or more TET enzymes comprise TET2. TET2 may be expressed and used as a fragment comprising TET2 residues 1129-1480 joined to TET2 residues 1844-1936 by a linker as described, e.g., in US Patent 10,961 ,525. In some embodiments, the one or more TET enzymes comprise TET1 and TET2. In some embodiments, the one or more TET enzymes comprise a V1900 TET mutant, such as a V1900A, V1900C, V1900G, V1900I, or V1900P TET mutant. In some embodiments, the one or more TET enzymes comprise a V1900 TET2 mutant, such as a V1900A, V1900C, V1900G, V1900I, or V1900P TET2 mutant. It can be beneficial to use a TET enzyme that maximizes formation of 5-carboxylcytosine (5-caC) relative to less oxidized modified cytosines, particularly 5- formylcytosine, because 5-caC is not a substrate for enzymatic deamination, e.g., by APOBEC enzymes such as APOBEC3A. Maximizing formation of 5-caC thus reduces the risk of false calls in which a base is identified as unmethylated because it underwent deamination even though it was methylated (or hydroxymethylated) in the original sample. Accordingly, in some embodiments, the TET enzyme comprises a mutation that increases formation of 5-caC. In some embodiments, the one or more TET enzymes comprise a TET2 enzyme comprising a T 1372S mutation, such as TET2-CS-T 1372S and TET2-CD- T1372S. A TET2 comprising a T1372S mutation is described in US Patent 10,961 ,525 and may be expressed and used as a fragment comprising TET2 residues 1129-1480 joined to TET2 residues 1844-1936 by a linker. Position 1372 of TET2 corresponds to position 258 of SEQ ID NO: 21 (wild type TET2 catalytic domain) of US Patent 10,961 ,525. Thus, the sequence of a T1372S TET2 catalytic domain may be obtained by changing the threonine at position 258 of SEQ ID NO: 21 of US Patent 10,961 ,525 to serine. TET2 comprising a T1372S mutation is also described in Liu et al., Nat Chem Biol. 2017 February; 13(2): 181-187. As demonstrated in Liu et al., TET2 comprising a T1372S mutation can more efficiently oxidize 5mC to produce 5-carboxylcytosine (5caC) than other versions of TET2 such as TET2 lacking a T1372S mutation. In some embodiments, the TET2 enzyme is a human TET2 enzyme comprising a T1372S mutation. Exemplary mutations are set forth above. “A mutation that increases formation of 5-caC” means that the TET enzyme having the mutation produces more 5-caC than a TET enzyme that lacks the mutation but is otherwise identical. 5-caC production can be measured as described, e.g., in Liu et al., Nat Chem Biol 13:181 -187 (2017) (see Online Methods section, TET reactions in vitro subsection, “driving” conditions). Any variants and/or mutants described in Liu et al. (2017) can be used in the disclosed methods as appropriate.
[00302] Provided herein is a method comprising contacting DNA contacting DNA with a mutant TET2 enzyme (e.g. comprising a V1900A, V1900C, V1900G, V1900I, V1900P, or T1372S mutation) to oxidize 5-methylcytosine (5mC) and/or 5- hydroxymethylcytosine (5hmC) present in the DNA to 5-carboxycytosine (5caC), subsequently contacting at least a portion of the DNA with a substituted borane reducing agent, thereby converting 5-caC in the DNA to dihydrouracil (DHU), thereby producing treated DNA, and sequencing at least a portion of the treated DNA.
[00303] In some embodiments, the procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA comprises separating DNA originally comprising the first nucleobase from DNA not originally comprising the first nucleobase. In some such embodiments, the first nucleobase is hmC. DNA originally comprising the first nucleobase may be separated from other DNA using a labeling procedure comprising biotinylating positions that originally comprised the first nucleobase. In some embodiments, the first nucleobase is first derivatized with an azide- containing moiety, such as a glucosyl-azide containing moiety. The azide-containing moiety then may serve as a reagent for attaching biotin, e.g., through Huisgen cycloaddition chemistry. Then, the DNA originally comprising the first nucleobase, now biotinylated, can be separated from DNA not originally comprising the first nucleobase using a biotin-binding agent, such as avidin, neutravidin (deglycosylated avidin with an isoelectric point of about 6.3), or streptavidin. An example of a procedure for separating DNA originally comprising the first nucleobase from DNA not originally comprising the first nucleobase is hmC-seal, which labels hmC to form [3-6-azide-glucosyl-5- hydroxymethylcytosine and then attaches a biotin moiety through Huisgen cycloaddition, followed by separation of the biotinylated DNA from other DNA using a biotin-binding agent. For an exemplary description of hmC-seal, see, e.g., Han et al., Mol. Cell 2016; 63: 711 -719. This approach is useful for identifying fragments that include one or more hmC nucleobases.
[00304] In some embodiments, following such a separation, the method further comprises differentially tagging each of the DNA originally comprising the first nucleobase, the DNA not originally comprising the first nucleobase. The method may further comprise pooling the DNA originally comprising the first nucleobase and the DNA not originally comprising the first nucleobase following differential tagging. The DNA originally comprising the first nucleobase and the DNA not originally comprising the first nucleobase may then be used in downstream analyses. For example, the pooled DNA originally comprising the first nucleobase and the DNA not originally comprising the first nucleobase may be sequenced in the same sequencing cell (such as after being subjected to further treatments, such as those described herein) while retaining the ability to resolve whether a given read came from a molecule of DNA originally comprising the first nucleobase or DNA not originally comprising the first nucleobase using the differential tags.
[00305] In some embodiments, the first nucleobase is a modified or unmodified adenine, and the second nucleobase is a modified or unmodified adenine. In some embodiments, the modified adenine is N6-methyladenine (mA). In some embodiments, the modified adenine is one or more of N6-methyladenine (mA), N6- hydroxymethyladenine (hmA), or N6-formyladenine (fA).
[00306] Techniques comprising partitioning based on methylation status or methylated DNA immunoprecipitation (MeDIP) can be used to separate DNA containing modified bases such as mC, mA, caC (which may be generated by oxidation of mC or hmC with Tet2, e.g., before enzymatic conversion of unmodified C to II, e.g., using a deaminase such as APOBEC3A), or dihydrouracil from other DNA. See, e.g., Kumar et al., Frontiers Genet. 2018; 9: 640; Greer et al., Cell 2015; 161 : 868-878. An antibody specific for mA is described in Sun et al., Bioessays 2015; 37:1155-62. Antibodies for various modified nucleobases, such as mC, caC, and forms of thymine/uracil including dihydrouracil or halogenated forms such as 5-bromouracil, are commercially available. Various modified bases can also be detected based on alterations in their base pairing specificity. For example, hypoxanthine is a modified form of adenine that can result from deamination and is read in sequencing as a G. See, e.g., US Patent 8,486,630; Brown, Genomes, 2nd Ed., John Wiley & Sons, Inc., New York, N.Y., 2002, chapter 14, “Mutation, Repair, and Recombination.”
G. Captured Set; Target Regions
[00307] In some embodiments, nucleic acids captured or enriched using a method described herein comprise captured DNA, such as one or more captured sets of DNA. In some embodiments, the captured DNA comprise target regions that are differentially methylated in different immune cell types. In some embodiments, the immune cell types comprise rare or closely related immune cell types, such as activated and naive lymphocytes or myeloid cells at different stages of differentiation.
[00308] In some embodiments, a captured epigenetic target region set captured from a sample or a subsample comprises genomic regions that show no or negligible methylation signal when analyzing cell-free DNA (cfDNA) from healthy individuals (e.g. in blood) but exhibit detectable methylation when analyzing cfDNA from individuals with cancer. Such regions are characterized by low background methylation levels in healthy populations, thereby providing an enhanced contrast that facilitates sensitive detection of tumor-derived DNA. In some embodiments, a captured epigenetic target region set captured from a sample or a subsample comprises genomic regions that show no or negligible methylation signal for a particular cell or tissue type (e.g., lung tissue) when analyzing cell-free DNA (cfDNA) from healthy individuals (e.g. in blood) but exhibit detectable methylation when analyzing cfDNA from individuals with disease associated with that particular cell or tissue type (e.g., if the tissue type is lung, then the disease can be lung cancer or pulmonary disorder).
[00309] In some embodiments, a captured epigenetic target region set captured from a sample or first subsample comprises hypermethylation target regions. In some embodiments, the hypermethylation target regions are differentially or exclusively hypermethylated in one cell type or in one immune cell type, or in one immune cell type within a cluster. In some embodiments, the hypermethylation target regions are hypermethylated to an extent that is distinguishably higher or exclusively present in one cell type or one immune cell type or one immune cell type within a cluster. Such hypermethylation target regions may be hypermethylated in other cell types but not to the extent observed in the one cell type. In some embodiments, the hypermethylation target regions show lower methylation in healthy cfDNA than in at least one other tissue type. [00310] In some embodiments, a captured epigenetic target region set captured from a sample or second subsample comprises hypomethylation target regions. In some embodiments, the hypomethylation target regions are exclusively hypomethylated in one cell type or in one immune cell type or in one immune cell type within a cluster. In some embodiments, the hypomethylation target regions are hypomethylated to an extent that is exclusively present in one cell type or one immune cell type or in one immune cell type within a cluster.
[00311] Such hypomethylation target regions may be hypomethylated in other cell types but not to the extent observed in the one cell type. In some embodiments, the hypomethylation target regions show higher methylation in healthy cfDNA than in at least one other tissue type. [0248] Without wishing to be bound by any particular theory, in an individual with cancer, proliferating or activated immune cells (and potentially also cancer cells) may shed more DNA into the bloodstream than immune cells in a healthy individual (and healthy cells of the same tissue type, respectively). As such, the distribution of cell type and/or tissue of origin of cfDNA may change upon carcinogenesis. For example, the distribution of immune cell type of origin may change in a subject having cancer, precancer, infection, transplant rejection, or other disease or disorder directly or indirectly affecting the immune system. The status of epigenetic target regions of certain immune cell types likewise may change in a subject having such a disease relative to a healthy subject or relative to the same subject prior to having the disease or disorder. Thus, variations in hypermethylation and/or hypomethylation can be an indicator of disease. For example, an increase in the level of hypermethylation target regions and/or hypomethylation target regions in a subsample following a partitioning step can be an indicator of the presence (or recurrence, depending on the history of the subject) of cancer.
[00312] Exemplary hypermethylation target regions and hypomethylation target regions useful for distinguishing between various cell types, including but not limited to immune cell types, have been identified by analyzing DNA obtained from various cell types via whole genome bisulfite sequencing, as described, e.g., in Stunnenberg, H. G. et. al., “The International Human Epigenome Consortium: A Blueprint for Scientific Collaboration and Discovery,” Cell 167, 1145 (2016) (doi.org/10.1186/sl3059-020-02065- 5). Whole-genome bisulfite sequencing data is available from the Blueprint consortium, available on the internet at dcc.blueprint-epigenome.eu.
[00313] In some embodiments, first and second captured target region sets comprise, respectively, DNA corresponding to a sequence-variable target region set and DNA corresponding to an epigenetic target region set, for example, as described in WO 2020/160414. The first and second captured sets may be combined to provide a combined captured set.
[00314] Where DNA (e.g., a sample or subsample) has been subjected to a procedure such as bisulfite conversion, treatment with a deaminase, or any of the other such procedures mentioned herein that alter the base-pairing specificity of certain bases, enrichment or capture may use oligonucleotides (e.g., primers or probes) specific for the altered or unaltered sequence, as desired.
[00315] In some embodiments in which a captured set comprising DNA corresponding to the sequence-variable target region set and the epigenetic target region set includes a combined captured set as discussed above, the DNA corresponding to the sequence-variable target region set may be present at a greater concentration than the DNA corresponding to the epigenetic target region set, e.g., a 1.1 to 1.2-fold greater concentration, a 1.2- to 1.4-fold greater concentration, a 1.4- to 1.6-fold greater concentration, a 1.6- to 1.8-fold greater concentration, a 1.8- to 2.0-fold greater concentration, a 2.0- to 2.2-fold greater concentration, a 2.2- to 2.4-fold greater concentration a 2.4- to 2.6-fold greater concentration, a 2.6- to 2.8-fold greater concentration, a 2.8- to 3.0-fold greater concentration, a 3.0- to 3.5-fold greater concentration, a 3.5- to 4.0, a 4.0- to 4.5-fold greater concentration, a 4.5- to 5.0-fold greater concentration, a 5.0- to 5.5-fold greater concentration, a 5.5- to 6.0-fold greater concentration, a 6.0- to 6.5-fold greater concentration, a 6.5- to 7.0-fold greater, a 7.0- to 7.5-fold greater concentration, a 7.5- to 8.0-fold greater concentration, an 8.0- to 8.5-fold greater concentration, an 8.5- to 9.0-fold greater concentration, a 9.0- to 9.5-fold greater concentration, 9.5- to 10.0-fold greater concentration, a 10- to 11 -fold greater concentration, an 11 - to 12-fold greater concentration a 12- to 13-fold greater concentration, a 13- to 14-fold greater concentration, a 14- to 15-fold greater concentration, a 15- to 16-fold greater concentration, a 16- to 17-fold greater concentration, a 17- to 18-fold greater concentration, an 18- to 19-fold greater concentration, a 19- to 20-fold greater concentration, a 20- to 30-fold greater concentration, a 30- to 40-fold greater concentration, a 40- to 50-fold greater concentration, a 50- to 60-fold greater concentration, a 60- to 70-fold greater concentration, a 70- to 80-fold greater concentration, a 80- to 90-fold greater concentration, or a 90- to 100-fold greater concentration. The degree of difference in concentrations accounts for normalization for the footprint sizes of the target regions, as discussed in the definition section.
1. Epigenetic Target Region Set
[00316] In some embodiments, an epigenetic target region set may comprise one or more types of target regions likely to differentiate DNA from different immune cell types and other non- immune cell types and/or to differentiate neoplastic (e.g., tumor or cancer) cells and from healthy cells, e.g., non-neoplastic circulating cells. Exemplary types of such regions are discussed in detail herein. The epigenetic target region set may also comprise one or more control regions, e.g., as described herein.
[00317] In some embodiments, the epigenetic target region set has a footprint of at least 100 kb, e.g., at least 200 kb, at least 300 kb, or at least 400 kb. In some embodiments, the epigenetic target region set has a footprint in the range of 100-1000 kb, e.g., 100-200 kb, 200-300 kb, 300- 400 kb, 400-500 kb, 500-600 kb, 600-700 kb, 700- 800 kb, 800-900 kb, and 900-1 ,000 kb. a. Hypermethylation Target Regions
[00318] In some embodiments, the epigenetic target region set comprises one or more hypermethylation target regions. In some embodiments, hypermethylation target regions are exclusively hypermethylated in one immune cell type or hypermethylated to a greater extent in one immune cell type than in any other immune cell type or than in any other immune cell type within the same immune cell cluster. In some such embodiments, hypermethylation target regions indicate the levels of particular immune cell types from which the DNA originated, including rare immune cell types such as activated B cells (including memory B cells and plasma cells), activated T cells (including regulatory T cells (Tregs), CD4 effector memory T cells, CD4 central memory T cells, CD8 effector memory T cells, and CD8 central memory T cells), and natural killer (NK) cells. Methylation patterns of hypermethylation target regions that are useful for deconvoluting immune cell types may further change in certain disease states, such as cancer. Thus, in some embodiments, hypermethylation target regions that are useful for deconvoluting immune cell types are also useful for determining the likelihood that the subject from which the sample was obtained has cancer or precancer. In some such embodiments, hypermethylation target regions are useful for determining whether levels of particular immune cell types are abnormal and whether such abnormal levels are likely related to the presence of cancer or precancer, or if they are related to a different disease or condition other than cancer or precancer.
[00319] In some embodiments, certain hypermethylation target regions exhibit an increase in the level of observed methylation, e.g., are hypermethylated, in DNA produced by neoplastic cells, such as tumor or cancer cells. Detection of such hypermethylation target regions, e.g., in conjunction with detection of hypermethylation target regions indicative of immune cell types, may further increase the specificity and/or sensitivity of methods described herein. In some embodiments, such increases in observed methylation in hypermethylated target regions indicate an increased likelihood that a sample (e.g., of cfDNA) was obtained from a subject having cancer. For example, hypermethylation of promoters of tumor suppressor genes has been observed repeatedly. See, e.g., Kang et ah, Genome Biol. 18:53 (2017) and references cited therein. In another example, as discussed above, hypermethylation target regions can include regions that do not necessarily differ in methylation in cancerous tissue relative to DNA from healthy tissue of the same type, but do differ in methylation (e.g., have more methylation) relative to cfDNA that is typical in healthy subjects. Where, for example, the presence of a cancer results in increased cell death such as apoptosis of cells of the tissue type corresponding to the cancer, such a cancer can be detected at least in part using such hypermethylation target regions. In some embodiments, hypermethylation target regions useful for determining the likelihood that a subject has cancer are different than the hypermethylation target regions useful for determining the levels of particular immune cell types. In some embodiments, at least some of the hypermethylation target regions useful for determining the likelihood that a subject has cancer are the same as the hypermethylation target regions useful for determining the levels of particular immune cell types.
[00320] An extensive discussion of methylation variable target regions in colorectal cancer is provided in Lam et al., Biochim Biophys Acta. 1866:106-20 (2016). These include VIM, SEPT9, ITGA4, OSM4, GATA4 and NDRG4. An exemplary set of hypermethylation target regions based on colorectal cancer (CRC) studies is provided in Table 1 . Many of these genes likely have relevance to cancers beyond colorectal cancer; for example, TP53 is widely recognized as a critically important tumor suppressor and hypermethylation-based inactivation of this gene may be a common oncogenic mechanism.
Table 1. Exemplary Hypermethylation Target Regions based on CRC studies.
[00321] In some embodiments, the hypermethylation target regions comprise a plurality of loci listed in Table 1 , e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the loci listed in Table 1. For example, for each locus included as a target region, there may be one or more probes with a hybridization site that binds between the transcription start site and the stop codon (the last stop codon for genes that are alternatively spliced) of the gene, or in the promoter region of the gene. In some embodiments, the one or more probes bind within 300 bp of the transcription start site of a gene in Table 1 , e.g., within 200 or 100 bp.
[00322] Methylation variable target regions in various types of lung cancer are discussed in detail, e.g., in Ooki et al., Clin. Cancer Res. 23:7141 -52 (2017); Belinksy, Annu. Rev. Physiol. 77:453- 74 (2015); Hulbert et al., Clin. Cancer Res. 23:1998-2005 (2017); Shi et al., BMC Genomics 18:901 (2017); Schneider et al., BMC Cancer. 11 :102 (2011 ); Lissa et al., Transl Lung Cancer Res 5(5):492-504 (2016); Skvortsova et al., Br. J. Cancer. 94(10): 1492-1495 (2006); Kim et al., Cancer Res. 61 :3419-3424 (2001 ); Furonaka et al., Pathology International 55:303-309 (2005); Gomes et al., Rev. Port. Pneumol. 20:20-30 (2014); Kim et al., Oncogene. 20:1765-70 (2001 ); Hopkins-Donaldson et al., Cell Death Differ. 10:356-64 (2003); Kikuchi et al., Clin. Cancer Res. 11 :2954-61 (2005); Heller et al., Oncogene 25:959-968 (2006); Licchesi et al., Carcinogenesis. 29:895-904 (2008); Guo et al., Clin. Cancer Res. 10:7917-24 (2004); Palmisano et al., Cancer Res. 63:4620-4625 (2003); and Toyooka et al., Cancer Res. 61 :4556-4560, (2001 ).
[00323] An exemplary set of hypermethylation target regions based on lung cancer studies is provided in Table 2. Many of these genes likely have relevance to cancers beyond lung cancer; for example, Casp8 (Caspase 8) is a key enzyme in programmed cell death and hypermethylation-based inactivation of this gene may be a common oncogenic mechanism not limited to lung cancer. Additionally, a number of genes appear in both Tables 1 and 2, indicating generality.
Table 2. Exemplary Hypermethylation Target Regions based on Lung Cancer studies.
[00324] Any of the foregoing embodiments concerning target regions identified in Table 2 may be combined with any of the embodiments described above concerning target regions identified in Table 1. In some embodiments, the hypermethylation target regions comprise a plurality of loci listed in Table 1 or Table 2, e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the loci listed in Table 1 or Table 2. [00325] In some embodiments, the hypermethylation target regions comprise regions of one or more genes listed in Table 2, e.g. at least 5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, 850, 900, 950, 1000, 1050,
[00326] 1000, 1100, 1150 or 1200 genes listed in Table 3. Hypermethylation of these genes can be useful for detecting contributions from immune cells to a DNA sample. In some embodiments, the hypermethylation target regions comprise regions of a plurality of genes listed in Table 2, e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the genes listed in Table 3. In some embodiments, the hypermethylation target regions comprise regions of all of the genes listed in Table 3.
Table 3. Exemplary genes comprising exemplary hypermethylation target regions.
[00327] Additional hypermethylation target regions may be obtained, e.g., from the Cancer Genome Atlas. Kang et al., Genome Biology 18:53 (2017), describe construction of a probabilistic method called CancerLocator using hypermethylation target regions from breast, colon, kidney, liver, and lung. In some embodiments, the hypermethylation target regions can be specific to one or more types of cancer. Accordingly, in some embodiments, the hypermethylation target regions include one, two, three, four, or five subsets of hypermethylation target regions that collectively show hypermethylation in one, two, three, four, or five of breast, colon, kidney, liver, and lung cancers.
[00328] In some embodiments, where different epigenetic target regions are captured from first and second subsamples, the epigenetic target regions captured from the first subsample comprise hypermethylation target regions. b. Hypomethylation target regions
[00329] In some embodiments, the epigenetic target region set comprises one or more hypomethylation target regions. In some embodiments, hypomethylation target regions are exclusively hypomethylated in one immune cell type or hypomethylated to a greater extent in one immune cell type than in any other immune cell type or in any other immune cell type within the same immune cell cluster. In some such embodiments, hypomethylation target regions indicate the levels of particular immune cell types from which the DNA originated, including rare immune cell types such as activated B cells (including memory B cells and plasma cells), activated T cells (including regulatory T cells (Tregs), CD4 effector memory T cells, CD4 central memory T cells, CD8 effector memory T cells, and CD8 central memory T cells), and natural killer (NK) cells. Methylation patterns of hypomethylation target regions that are useful for deconvoluting immune cell types may further change in certain disease states, such as cancer. Thus, in some embodiments, hypomethylation target regions that are useful for deconvoluting immune cell types are also useful for determining the likelihood that the subject from which the sample was obtained has cancer or precancer. In some such embodiments, hypomethylation target regions are useful for determining whether levels of particular immune cell types are abnormal and whether such abnormal levels are likely related to the presence of cancer or precancer, or if they are related to a different disease or condition other than cancer or precancer.
[00330] Additionally, global hypomethylation is a commonly observed phenomenon in various cancers. See, e.g., Hon et al., Genome Res. 22:246-258 (2012) (breast cancer); Ehrlich, Epigenomics 1 :239-259 (2009) (review article noting observations of hypomethylation in colon, ovarian, prostate, leukemia, hepatocellular, and cervical cancers). For example, regions such as repeated elements, e.g., LINE1 elements, Alu elements, centromeric tandem repeats, pericentromeric tandem repeats, and satellite DNA, and intergenic regions that are ordinarily methylated in healthy cells may show reduced methylation in tumor cells. Accordingly, in some embodiments, the epigenetic target region set includes hypomethylation target regions in which a decrease in the level of observed methylation indicates an increased likelihood of the presence of cancer. Detection of such hypomethylation target regions, e.g., in conjunction with detection of hypomethylation target regions indicative of immune cell types, may further increase the specificity and/or sensitivity of methods described herein. In another example, as discussed above, hypomethylation target regions can include regions that do not necessarily differ in methylation in cancerous tissue relative to DNA from healthy tissue of the same type, but do differ in methylation (e.g., are less methylated) relative to cfDNA that is typical in healthy subjects. Where, for example, the presence of a cancer results in increased cell death such as apoptosis of cells of the tissue type corresponding to the cancer, such a cancer can be detected at least in part using such hypomethylation target regions. In some embodiments, hypomethylation target regions useful for determining the likelihood that a subject has cancer are different than the hypomethylation target regions useful for determining the levels of particular immune cell types. In some embodiments, at least some of the hypomethylation target regions useful for determining the likelihood that a subject has cancer are the same as the hypomethylation variable target regions useful for determining the levels of particular immune cell types.
[00331] In some embodiments, hypomethylation target regions include repeated elements and/or intergenic regions. In some embodiments, repeated elements include one, two, three, four, or five of LINE1 elements, Alu elements, centromeric tandem repeats, pericentromeric tandem repeats, and/or satellite DNA.
[00332] Exemplary specific genomic regions that show cancer-associated hypomethylation include nucleotides 8403565-8953708 and 151104701-151106035 of human chromosome 1. In some embodiments, the hypomethylation target regions overlap or comprise one or both of these regions.
[00333] Additionally, hypomethylation target regions may be obtained, e.g., from Fox-Fisher et al., ElifeNov 29; 10 (2021 ), EpiDISH R package, Moss et al., Nat Commun 9:1 (2018), and Loyfer et al. bioRxiv https://doi.org/10.1101/2022.01.24.477547 (2022). In some embodiments, the hypomethylation target regions can be specific to one or more types of immune cells.
[00334] In some embodiments, where different epigenetic target regions are captured from first and second subsamples, the epigenetic target regions captured from the second subsample comprise hypomethylation target regions. In some embodiments, the epigenetic target regions captured from the second subsample comprise hypomethylation target regions and the epigenetic target regions captured from the first subsample comprise hypermethylation target regions. c. CTCF binding regions
[00335] CTCF is a DNA-binding protein that contributes to chromatin organization and often colocalizes with cohesin. Perturbation of CTCF binding sites has been reported in a variety of different cancers. See, e.g., Katainen et al., Nature Genetics, doi:10.1038/ng.3335, published online 8 June 2015; Guo et al., Nat. Commun. 9:1520 (2018). CTCF binding results in recognizable patterns in cfDNA that can be detected by sequencing, e.g., through fragment length analysis. Details regarding sequencing-based fragment length analysis are provided in Snyder et al., Cell 164:57-68 (2016); WO 2018/009723; and US20170211143A1 , each of which are incorporated herein by reference.
[00336] Thus, perturbations of CTCF binding result in variation in the fragmentation patterns of cfDNA. As such, CTCF binding sites are a type of fragmentation variable target regions.
[00337] There are many known CTCF binding sites. See, e.g., the CTCFBSDB (CTCF Binding Site Database), available on the Internet at insulatordb.uthsc.edu/; Cuddapah et al., Genome Res. 19:24-32 (2009); Martin et al., Nat. Struct. Mol. Biol. 18:708-14 (2011 ); Rhee et al., Cell. 147:1408-19 (2011 ), each of which are incorporated by reference. Exemplary CTCF binding sites are at nucleotides 56014955-56016161 on chromosome 8 and nucleotides 95359169-95360473 on chromosome 13.
[00338] Accordingly, in some embodiments, the epigenetic target region set includes CTCF binding regions. In some embodiments, the CTCF binding regions comprise at least 10, 20, 50, 100, 200, or 500 CTCF binding regions, or 10-20, 20-50, 50- 100, 100-200, 200-500, or 500-1000 CTCF binding regions, e.g., such as CTCF binding regions described above or in one or more of CTCFBSDB or the Cuddapah et al., Martin et al., or Rhee et al. articles cited above.
[00339] In some embodiments, at least some of the CTCF sites can be methylated or unmethylated, wherein the methylation state is correlated with the whether or not the cell is a cancer cell. In some embodiments, the epigenetic target region set comprises at least 100 bp, at least 200 bp, at least 300 bp, at least 400 bp, at least 500 bp, at least 750 bp, at least 1000 bp upstream and downstream regions of the CTCF binding sites. d. Transcription start sites.
[00340] T ranscription start sites may also show perturbations in neoplastic cells. For example, nucleosome organization at various transcription start sites in healthy cells of the hematopoietic lineage — which contributes substantially to cfDNA in healthy individuals — may differ from nucleosome organization at those transcription start sites in neoplastic cells. This results in different cfDNA patterns that can be detected by sequencing, as discussed generally in Snyder et al., Cell 164:57-68 (2016); WO 2018/009723; and US20170211143A1. In another example, transcription start sites may not necessarily differ epigenetically in cancerous tissue relative to DNA from healthy tissue of the same type, but do differ epigenetically (e.g., with respect to nucleosome organization) relative to cfDNA that is typical in healthy subjects. Where, for example, the presence of a cancer results in increased cell death, such as apoptosis, of cells of the tissue type corresponding to the cancer, such a cancer can be detected at least in part using such differences in transcription start sites.
[00341] Thus, perturbations of transcription start sites also result in variation in the fragmentation patterns of cfDNA. As such, transcription start sites are also a type of fragmentation variable target regions.
[00342] Human transcriptional start sites are available from DBTSS (DataBase of Human Transcription Start Sites), available on the Internet at dbtss.hgc.jp and described in Yamashita et al., Nucleic Acids Res. 34(Database issue): D86-D89 (2006), which is incorporated herein by reference.
[00343] Accordingly, in some embodiments, the epigenetic target region set includes transcriptional start sites. In some embodiments, the transcriptional start sites comprise at least 10, 20, 50, 100, 200, or 500 transcriptional start sites, or 10-20, 20-50, 50-100, 100-200, 200-500, or 500-1000 transcriptional start sites, e.g., such as transcriptional start sites listed in DBTSS. In some embodiments, at least some of the transcription start sites can be methylated or unmethylated, wherein the methylation state is correlated with whether or not the cell is a cancer cell. In some embodiments, the epigenetic target region set comprises at least 100 bp, at least 200 bp, at least 300 bp, at least 400 bp, at least 500 bp, at least 750 bp, at least 1000 bp upstream and downstream regions of the transcription start sites. e. Focal amplifications
[00344] Although focal amplifications are somatic mutations, they can be detected by sequencing based on read frequency in a manner analogous to approaches for detecting certain epigenetic changes such as changes in methylation. As such, regions that may show focal amplifications in cancer can be included in the epigenetic target region set and may comprise one or more of AR, BRAF, CCND1 , CCND2, CCNE1 , CDK4, CDK6, EGFR, ERBB2, FGFR1 , FGFR2, KIT, KRAS, MET, MYC, PDGFRA, PIK3CA, and RAF1 . For example, in some embodiments, the epigenetic target region set comprises at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 , 12, 13, 1 , 15, 16, 17, or 18 of the foregoing targets. f. Methylation control regions or control regions
[00345] It can be useful to include control regions to facilitate data validation. In some embodiments, the epigenetic target region set includes control regions that are expected to be methylated or unmethylated in essentially all samples, regardless of whether the DNA is derived from a cancer cell or a normal cell. In some embodiments, the epigenetic target region set includes negative control regions that are expected to be hypomethylated or unmethylated in essentially all samples. In some embodiments, the epigenetic target region set includes positive control regions that are expected to be hypermethylated in essentially all samples.
2. Sequence-variable target region set.
[00346] In some embodiments, the sequence-variable target region set comprises a plurality of regions known to undergo somatic mutations (e.g., single nucleotide variations and/or indels) in cancer. The single nucleotide variations and/or indels may be relative to a reference sequence, e.g., a published human genome sequence, such as the GRCh38 human genome assembly.
[00347] In some aspects, the sequence-variable target region set targets a plurality of different genes or genomic regions (“panel”) selected such that a determined proportion of subjects having a cancer exhibits a genetic variant or tumor marker in one or more different genes or genomic regions in the panel. The panel may be selected to limit a region for sequencing to a fixed number of base pairs. The panel may be selected to sequence a desired amount of DNA, e.g., by adjusting the affinity and/or amount of the probes as described elsewhere herein. The panel may be further selected to achieve a desired sequence read depth. The panel may be selected to achieve a desired sequence read depth or sequence read coverage for an amount of sequenced base pairs. The panel may be selected to achieve a theoretical sensitivity, a theoretical specificity, and/or a theoretical accuracy for detecting one or more genetic variants in a sample. [0284] Probes for detecting the panel of regions can include those for detecting genomic regions of interest (hotspot regions). Information about chromatin structure can be taken into account in designing probes, and/or probes can be designed to maximize the likelihood that particular sites (e.g., KRAS codons 12 and 13) can be captured, and may be designed to optimize capture based on analysis of cfDNA coverage and fragment size variation impacted by nucleosome binding patterns and GC sequence composition. Regions used herein can also include non-hotspot regions optimized based on nucleosome positions and GC models.
[00348] Examples of listings of genomic locations of interest may be found in Table 4 and Table 5. In some embodiments, a sequence-variable target region set used in the methods of the present disclosure comprises at least a portion of at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, or 70 of the genes of Table 3. In some embodiments, a sequence-variable target region set used in the methods of the present disclosure comprises at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, or 70 of the SNVs of Table 4. In some embodiments, a sequence-variable target region set used in the methods of the present disclosure comprises at least 1 , at least 2, at least 3, at least 4, at least 5, or 6 of the fusions of Table 4. In some embodiments, a sequencevariable target region set used in the methods of the present disclosure comprise at least a portion of at least 1 , at least 2, or 3 of the indels of Table 4. In some embodiments, a sequence-variable target region set used in the methods of the present disclosure comprises at least a portion of at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, at least 70, or 73 of the genes of Table 5. In some embodiments, a sequence- variable target region set used in the methods of the present disclosure comprises at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, at least 70, or 73 of the SNVs of Table 5. In some embodiments, a sequence-variable target region set used in the methods of the present disclosure comprises at least 1 , at least 2, at least 3, at least 4, at least 5, or 6 of the fusions of Table 5. In some embodiments, a sequence-variable target region set used in the methods of the present disclosure comprises at least a portion of at least 1 , at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11 , at least 12, at least 13, at least 14, at least 15, at least 16, at least 17, or 18 of the indels of Table 5. Each of these genomic locations of interest may be identified as a backbone region or hot-spot region for a given panel. An example of a listing of hot-spot genomic locations of interest may be found in Table 6. In some embodiments, a sequence-variable target region set used in the methods of the present disclosure comprises at least a portion of at least 1 , at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11 , at least 12, at least 13, at least 14, at least 15, at least 16, at least 17, at least 18, at least 19, or at least 20 of the genes of Table 6. Each hot-spot genomic region is listed with several characteristics, including the associated gene, chromosome on which it resides, the start and stop position of the genome representing the gene’s locus, the length of the gene’s locus in base pairs, the exons covered by the gene, and the critical feature (e g., type of mutation) that a given genomic region of interest may seek to capture.
Table 4
Table 5
Table 6
[00349] Additionally, or alternatively, suitable target region sets are available from the literature. For example, Gale et al., PLoS One 13: e0194630 (2018), which is incorporated herein by reference, describes a panel of 35 cancer-related gene targets that can be used as part or all of a sequence-variable target region set. These 35 targets are AKTI, ALK, BRAF, CCND1 , CDK2A, CTNNB1 , EGFR, ERBB2, ESR1 , FGFR1 , FGFR2, FGFR3, F0XL2, GAT A3, GNA11 , GNAQ, GNAS, HRAS, IDH1 , IDH2, KIT, KRAS, MED 12, MET, MYC, NFE2L2, NRAS, PDGFRA, PIK3CA, PPP2R1A, PTEN, RET, STK11 , TP53, and U2AF1 .
[00350] In some embodiments, the sequence-variable target region set comprises target regions from at least 10, 20, 30, or 35 cancer-related genes, such as the cancer- related genes listed above.
H. Subjects
[00351] In some embodiments, the DNA (e.g., cfDNA) is obtained from a subject having a cancer or a precancer, an infection, transplant rejection, or other disease directly or indirectly affecting the immune system. In some embodiments, the DNA (e.g., cfDNA) is obtained from a subject suspected of having a cancer or a precancer, an infection, transplant rejection, or other disease directly or indirectly affecting the immune system. In some embodiments, the DNA (e.g., cfDNA) is obtained from a subject having a tumor. In some embodiments, the DNA (e.g., cfDNA) is obtained from a subject suspected of having a tumor. In some embodiments, the DNA (e.g., cfDNA) is obtained from a subject having neoplasia. In some embodiments, the DNA (e.g., cfDNA) is obtained from a subject suspected of having neoplasia. In some embodiments, the DNA (e.g., cfDNA) is obtained from a subject in remission from a tumor, cancer, or neoplasia (e.g., following chemotherapy, surgical resection, radiation, or a combination thereof). In any of the foregoing embodiments, the cancer, tumor, or neoplasia or suspected cancer, tumor, or neoplasia may be of the lung, colon, rectum, kidney, breast, prostate, or liver. In some embodiments, the cancer, tumor, or neoplasia or suspected cancer, tumor, or neoplasia is of the lung. In some embodiments, the cancer, tumor, or neoplasia or suspected cancer, tumor, or neoplasia is of the colon or rectum. In some embodiments, the cancer, tumor, or neoplasia or suspected cancer, tumor, or neoplasia is of the breast. In some embodiments, the cancer, tumor, or neoplasia or suspected cancer, tumor, or neoplasia is of the prostate. In any of the foregoing embodiments, the subject may be a human subject. I. Pooling of DNA from samples or subsamples or portions thereof
[00352] In some embodiments, the methods herein comprise preparing one or more pools comprising tagged DNA from a plurality of partitioned subsamples. In some embodiments, a pool comprises at least a portion of the DNA of a hypomethylated partition and at least a portion of the DNA of a hypermethylated partition. Target regions, e.g., including epigenetic target regions and/or sequence-variable target regions, may be captured from a pool. The steps of capturing a target region set from at least an aliquot or portion of a sample or subsample described elsewhere herein encompass capture steps performed on a pool comprising DNA from first and second subsamples. A step of amplifying DNA in a pool may be performed before capturing target regions from the pool. The capturing step may have any of the features described for capturing steps elsewhere herein.
[00353] In some embodiments, the methods comprise preparing a first pool comprising at least a portion of the DNA of a hypomethylated partition. In some embodiments, the methods comprise preparing a second pool comprising at least a portion of the DNA of a hypermethylated partition. In some embodiments, the methods comprise capturing at least a first set of target regions from the first pool, wherein the first set comprises sequence-variable target regions. A step of amplifying DNA in the first pool may be performed before this capture step. In some embodiments, capturing the first set of target regions from the first pool comprises contacting the DNA of the first pool with a first set of target-specific probes, wherein the first set of target- specific probes comprises target-binding probes specific for the sequence-variable target regions. In some embodiments, the methods comprise capturing a second plurality of sets of target regions from the second pool, wherein the second plurality comprises sequence-variable target regions and epigenetic target regions. A step of amplifying DNA in the second pool may be performed before this capture step. In some embodiments, capturing the second plurality of sets of target regions from the second pool comprises contacting the DNA of the first pool with a second set of target-specific probes, wherein the second set of targetspecific probes comprises target-binding probes specific for the sequence-variable target regions and target-binding probes specific for the epigenetic target regions. [00354] In some embodiments, sequence-variable target regions are captured from a second portion of a partitioned subsample. The second portion may include some, a majority, substantially all, or all of the DNA of the subsample that was not included in the pool. The regions captured from the pool and from the subsample may be combined and analyzed in parallel.
[00355] The epigenetic target regions may show differences in methylation levels and/or fragmentation patterns depending on whether they originated from a particular cell or tissue type or from a tumor or from healthy cells, as discussed elsewhere herein. The sequence-variable target regions may show differences in sequence depending on whether they originated from a tumor or from healthy cells. [0293] Analysis of epigenetic target regions from a hypomethylated partition may be less informative in some applications than analysis of sequence-variable target regions from hypermethylated and hypomethylated partitions and epigenetic target regions from a hypermethylated partition. As such, in methods where sequence-variable target regions and epigenetic target regions are being captured, the latter may be captured to a lesser extent than one or more of the sequence-variable target regions are captured from the hypermethylated and hypomethylated partitions and/or to a lesser extent that epigenetic target regions are captured from a hypermethylated partition. For example, sequence-variable target regions can be captured from a portion of a hypomethylated partition that is not pooled with a hypermethylated partition, and the pool can be prepared with some (e g., a majority, substantially all, or all) of the DNA from a hypermethylated partition and none or some (e.g., a minority) of the DNA from a hypomethylated partition. Such approaches can reduce or eliminate sequencing of epigenetic target regions from hypomethylated partitions, thereby reducing the amount of sequencing data that suffices for further analysis.
[00356] In some embodiments, including a minority of the DNA of a hypomethylated partition in the pool facilitates quantification of one or more epigenetic features (e.g., methylation or other epigenetic feature(s) discussed in detail elsewhere herein), e.g., on a relative basis.
[00357] In some embodiments, the pool comprises a minority of the DNA of a hypomethylated partition, e.g., less than about 50% of the DNA of a hypomethylated partition, such as less than or equal to about 45%, 40%, 35%, 30%, 25%, 20%, 15%, 10%, or 5% of the DNA of a hypom ethylated partition. In some embodiments, the pool comprises about 5%-25% of the DNA of a hypomethylated partition. In some embodiments, the pool comprises about 10%-20% of the DNA of a hypomethylated partition. In some embodiments, the pool comprises about 10% of the DNA of a hypomethylated partition. In some embodiments, the pool comprises about 15% of the DNA of a hypomethylated partition. In some embodiments, the pool comprises about 20% of the DNA of a hypomethylated partition.
[00358] In some embodiments, the pool comprises a portion of a hypermethylated partition, which may be at least about 50% of the DNA of a hypermethylated partition. For example, the pool may comprise at least about 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, or 95% of the DNA of a hypermethylated partition. In some embodiments, the pool comprises 50-55%, 55- 60%, 60-65%, 65-70%, 70-75%, 75-80%, 80-85%, 85-90%, 90- 95%, or 95-100% of the DNA of a hypermethylated partition. In some embodiments, the second pool comprises all or substantially all of the DNA of a hypermethylated partition.
[00359] In some embodiments, a first pool comprises substantially all or all of the DNA of a hypomethylated partition (e.g., wherein a second pool does not comprise DNA of a hypomethylated partition. In some embodiments, the second pool does not comprise DNA of a hypomethylated partition (e.g., wherein the first pool comprises substantially all or all of the DNA of a hypomethylated partition).
[00360] In some embodiments, a second pool comprises a portion of a hypermethylated partition, which may be any of the values and ranges set forth above with respect to a hypomethylated partition. In some embodiments, the second pool comprises all or substantially all of the DNA of a hypermethylated partition.
[00361] In an exemplary embodiment, after partitioning, the partitions separately undergo end repair and ligation to adapters comprising molecular barcodes and are then amplified separately. After the amplification, amplified molecules are enriched (still keeping the partitions separate). Post-enrichment, the enriched DNA are pooled according to any of the embodiments described herein, and then amplified again. After amplification, the molecules are sequenced. [00362] In various embodiments, the methods further comprise sequencing the captured DNA, e.g., to different degrees of sequencing depth for the epigenetic and sequence-variable target region sets, consistent with the discussion above.
J. Sequencing
[00363] In general, sample nucleic acids, including nucleic acids flanked by adapters, with or without prior amplification can be subject to sequencing. Sequencing methods include, for example, Sanger sequencing, high-throughput sequencing, pyrosequencing, sequencing-by synthesis, single-molecule sequencing, nanopore sequencing, semiconductor sequencing, sequencing-by-ligation, sequencing-by- hybridization, Digital Gene Expression (Helicos), Next generation sequencing (NGS), Single Molecule Sequencing by Synthesis (SMSS) (Helicos), massively-parallel sequencing, Clonal Single Molecule Array (Solexa), shotgun sequencing, Ion Torrent, Oxford Nanopore, Roche Genia, Maxim-Gilbert sequencing, primer walking, and sequencing using PacBio, SOLiD, Ion Torrent, or Nanopore platforms.
[00364] In some embodiments, sequencing comprises detecting and/or distinguishing unmodified and modified nucleobases. For example, PacBio sequencing (e.g., single-molecule real-time (SMRT) sequencing) offers the ability to directly detect of, e.g., 5-methylcytosine and 5- hydroxymethylcytosine as well as unmodified cytosine. See, e.g., Schatz., Nature Methods. 14(4): 347-348 (2017); and US 9,150,918. Also, Oxford nanopore sequencing systems (e.g., MinlON sequencer) that can directly detect methylation of DNA (for example: 5-methylcytosine and 5-hydroxymethylcytosine) can be used here. Sequencing reactions can be performed in a variety of sample processing units, which may multiple lanes, multiple channels, multiple wells, or other mean of processing multiple sample sets substantially simultaneously. Sample processing unit can also include multiple sample chambers to enable processing of multiple runs simultaneously. Similarly, Ion Torrent sequencing may also be used to directly detect methylation. Thus, in some embodiments, methylation status can be determined during sequencing, e.g., without or independently of a partitioning step or a conversion procedure such as bisulfite treatment. [00365] The sequencing reactions can be performed on one or more forms of nucleic acids, such as those known to contain markers of cancer or of other disease. The sequencing reactions can also be performed on any nucleic acid fragments present in the sample. In some embodiments, sequence coverage of the genome may be less than 5%, 10%, 15%, 20%, 25%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, 99%, 99.9% or 100%. In some embodiments, the sequence reactions may provide for sequence coverage of at least 5%, 10%, 15%, 20%, 25%, 30%, 40%, 50%, 60%, 70%, or 80% of the genome. Sequence coverage can be performed on at least 5, 10, 20, 70, 100, 200 or 500 different genes, or at most 5000, 2500, 1000, 500 or 100 different genes. [0304] Simultaneous sequencing reactions may be performed using multiplex sequencing. In some cases, cell-free nucleic acids may be sequenced with at least 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 50000, 100,000 sequencing reactions. In other cases, cell-free nucleic acids may be sequenced with less than 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 50000, 100,000 sequencing reactions. Sequencing reactions may be performed sequentially or simultaneously. Subsequent data analysis may be performed on all or part of the sequencing reactions. In some cases, data analysis may be performed on at least 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 50000, 100,000 sequencing reactions. In other cases, data analysis may be performed on less than 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 50000, 100,000 sequencing reactions. An exemplary read depth is 1000- 50000 reads per locus (base). 1.
1. Differential depth of sequencing
[00366] In some embodiments, nucleic acids corresponding to a sequence-variable target region set are sequenced to a greater depth of sequencing than nucleic acids corresponding to an epigenetic target region set. For example, the depth of sequencing for nucleic acids corresponding to sequence variant target region sets may be at least 1.25-, 1.5-, 1.75-, 2-, 2.25-, 2.5-, 2.75-, 3-, 3.5-, 4-, 4.5-, 5-, 6-, 7-, 8-, 9-, 10-, 11 -, 12-, 13- , 14-, or 15 -fold greater, or 1.25- to 1.5-, 1.5- to 1.75-, 1.75- to 2-, 2- to 2.25-, 2.25- to 2.5-, 2.5- to 2.75-, 2.75- to 3-, 3- to 3.5-, 3.5- to 4-, 4- to 4.5-, 4.5- to 5-, 5- to 5.5-, 5.5- to 6-, 6- to 7-, 7- to 8-, 8- to 9-, 9- to 10-, 10- to 11 -, 11 - to 12-, 13- to 14-, 14- to 15-fold, or 15- to 100-fold greater, than the depth of sequencing for nucleic acids corresponding to an epigenetic target region set. In some embodiments, said depth of sequencing is at least 2-fold greater. In some embodiments, said depth of sequencing is at least 5-fold greater. In some embodiments, said depth of sequencing is at least 10-fold greater. In some embodiments, said depth of sequencing is 4- to 10-fold greater. In some embodiments, said depth of sequencing is 4- to 100-fold greater.
[00367] In some embodiments, DNA corresponding to a sequence-variable target region set, and/or to an epigenetic target region set are sequenced concurrently, e.g., in the same sequencing cell (such as the flow cell of an Illumina sequencer) and/or in the same composition, which may be a combined or pooled composition resulting from recombining separately captured sets or a composition obtained by, e.g., capturing the cfDNA corresponding to the sequence-variable target region set, and/or the captured cfDNA corresponding to an epigenetic target region set in the same vessel.
K. Analysis
[00368] In some embodiments, any of the methods disclosed herein comprises determining a likelihood that the subject from which the DNA was obtained has a disease or disorder related to the immune system, such as an infection, transplant rejection, or cancer or precancer.
[00369] In some embodiments, any of the methods disclosed herein comprises identifying the presence of DNA produced by a tumor (or neoplastic cells, or cancer cells) or by precancer cells. In some embodiments, a method described herein comprises determining an indication of cancer in the subject. In some such embodiments, determination of the indication of cancer facilitates detection or diagnosis or cancer or precancer, or determination of cancer prognosis or cancer treatment options. For example, determining the metrics from the one or more classification regions and the one or more control regions can help in determining the indication of cancer. In some embodiments, the metrics can be used to determine the tumor fraction of a sample.
[00370] The present methods can be used to diagnose presence of conditions, particularly cancer or precancer, in a subject, to characterize conditions (e.g., staging cancer or determining heterogeneity of a cancer), monitor response to treatment of a condition, effect prognosis risk of developing a condition or subsequent course of a condition. The present disclosure can also be useful in determining the efficacy of a particular treatment option. For example, the change in the tumor fraction or determining the methylation status of one or regions can be useful in determining whether the patient is responding to the treatment or not. In another example, perhaps certain treatment options may be correlated with methylation profiles of cancers over time. This correlation may be useful in selecting a therapy.
[00371] Additionally, if a cancer is observed to be in remission after treatment, the present methods can be used to monitor residual disease or recurrence of disease.
[00372] The types and number of cancers that may be detected may include blood cancers, brain cancers, lung cancers, skin cancers, nose cancers, throat cancers, liver cancers, bone cancers, lymphomas, pancreatic cancers, skin cancers, bowel cancers, rectal cancers, thyroid cancers, bladder cancers, kidney cancers, mouth cancers, stomach cancers, solid state tumors, heterogeneous tumors, homogenous tumors and the like. Type and/or stage of cancer can be detected from genetic variations including mutations, rare mutations, indels, copy number variations, transversions, translocations, recombination, inversion, deletions, aneuploidy, partial aneuploidy, polyploidy, chromosomal instability, chromosomal structure alterations, gene fusions, chromosome fusions, gene truncations, gene amplification, gene duplications, chromosomal lesions, DNA lesions, abnormal changes in nucleic acid chemical modifications, abnormal changes in epigenetic patterns, and abnormal changes in nucleic acid 5-methylcytosine. [00373] Genetic data can also be used for characterizing a specific form of cancer. Cancers are often heterogeneous in both composition and staging. Genetic profile data may allow characterization of specific sub-types of cancer that may be important in the diagnosis or treatment of that specific sub-type. This information may also provide a subject or practitioner clues regarding the prognosis of a specific type of cancer and allow either a subject or practitioner to adapt treatment options in accord with the progress of the disease. Some cancers can progress to become more aggressive and genetically unstable. Other cancers may remain benign, inactive or dormant. The system and methods of this disclosure may be useful in determining disease progression.
[00374] Further, the methods of the disclosure may be used to characterize the heterogeneity of an abnormal condition in a subject. Such methods can include, e.g., generating a genetic profile of extracellular polynucleotides derived from the subject, wherein the genetic profile comprises a plurality of data resulting from copy number variation and rare mutation analyses. In some embodiments, an abnormal condition is cancer or precancer. In some embodiments, the abnormal condition may be one resulting in a heterogeneous genomic population. In the example of cancer, some tumors are known to comprise tumor cells in different stages of the cancer. In other examples, heterogeneity may comprise multiple foci of disease. Again, in the example of cancer, there may be multiple tumor foci, perhaps where one or more foci are the result of metastases that have spread from a primary site.
[00375] The present methods can be used to generate or profile, fingerprint or set of data that is a summation of genetic information derived from different cells in a heterogeneous disease. This set of data may comprise copy number variation, epigenetic variation, or other mutation analyses alone or in combination.
[00376] The present methods can be used to diagnose, prognose, monitor or observe cancers, or other diseases. In some embodiments, the methods herein do not involve the diagnosing, prognosing or monitoring a fetus and as such are not directed to non-invasive prenatal testing. In other embodiments, these methodologies may be employed in a pregnant subject to diagnose, prognose, monitor or observe cancers or other diseases in an unborn subject whose DNA and other polynucleotides may cocirculate with maternal molecules.
[00377] An exemplary method for determining an indication of cancer through NGS comprises the following steps:
1. Extracting cfDNA from a blood sample
2. Partitioning cfDNA into a plurality of partitions by contacting the DNA with an agent that recognizes a modified cytosine, such as methyl cytosine, in the DNA
3. Ligating the partitions with adapters comprising molecular barcodes
4. Treating the hyper and/or intermediate partitions with one or more MSREs and/or treating the hypo partition with one or more MDREs
5. Amplifying the partitions post digestion via PCR amplification
6. Capturing DNA comprising hypermethylated target regions, hypomethylated target regions and control regions using target-specific probes. 7. Amplifying the captured DNA and assaying in multiplex on an NGS instrument.
8. Analyzing NGS data using one or more methods disclosed herein to determine the indication of cancer.
[00378] Another exemplary method for determining an indication of cancer through NGS comprises the following steps:
1. Extracting cfDNA from a blood sample.
2. Ligating the cfDNA with adapters comprising molecular barcodes
3. Subjecting the ligated cfDNA to a methylation-sensitive conversion method
4. Capturing DNA comprising hypermethylated target regions, hypomethylated target regions and control regions using target-specific probes, wherein the probes are designed such that they can be targeted to capture converted or unconverted molecules depending on the type of methylation-sensitive conversion method and the target regions (whether hypermethylated target regions or hypomethylated target regions) being captured.
5. Amplifying the captured DNA and assaying in multiplex on an NGS instrument.
6. Analyzing NGS data using one or more methods disclosed herein to determine the indication of cancer.
[00379] Another exemplary method for determining methylation status of a target region (e.g., promoter region) through NGS comprises the following steps:
1. Extracting cfDNA from a blood sample
2. Partitioning cfDNA into a plurality of partitions by contacting the DNA with an agent that recognizes a modified cytosine, such as methyl cytosine, in the DNA
3. Ligating the partitions with adapters comprising molecular barcodes
4. Treating the hyper and/or intermediate partitions with one or more MSREs and/or treating the hypo partition with one or more MDREs
5. Amplifying the partitions post digestion via PCR amplification
6. Capturing DNA comprising hypermethylated target regions, hypomethylated target regions and control regions using target-specific probes.
7. Amplifying the captured DNA and assaying in multiplex on an NGS instrument.
8. Analyzing NGS data using one or more methods disclosed herein to determine whether the target region (e.g., promoter region) is methylated or not. [00380] Another exemplary method for determining an indication of cancer or for determining methylation status of a target region (e.g., promoter region) through NGS comprises the following steps:
1. Extracting cfDNA from a blood sample
2. Partitioning cfDNA into a plurality of partitions by contacting the DNA with an agent that recognizes a modified cytosine, such as methyl cytosine, in the DNA
3. Ligating the partitions with adapters comprising molecular barcodes
4. Treating the hyper and/or intermediate partitions with a procedure that affect a first nucleobase in the cfDNA differently from a second nucleobase (e.g., bisulfite method, EM-seq)
5. Amplifying the partitions post -treating the procedure via PCR amplification
6. Capturing DNA comprising hypermethylated target regions, hypomethylated target regions and control regions using target-specific probes.
7. Amplifying the captured DNA and assaying in multiplex on an NGS instrument.
8. Analyzing NGS data using one or more methods disclosed herein to determine whether the target region (e.g., promoter region) is methylated or not.
[00381] In some embodiments, instead of using cfDNA from a blood sample, the exemplary methods discussed above can also be used with DNA samples obtained from tissue sample, stool sample or bodily fluids like urine sample. In these embodiments, the DNA can be a whole genomic DNA. In instances, where whole genomic DNA is used, an additional step of fragmenting the DNA (after DNA extraction, but prior to step 2) is performed to the methods discussed above. In some embodiments of methods described herein, molecular barcodes consist of nucleotides that are not altered by a procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA, such as any of those described herein (e.g., mC along with A, T, and G where the procedure is bisulfite conversion or any other conversion that does not affect mC; hmC along with A, T, and G where the procedure is a conversion that does not affect hmC; etc.). In some embodiments of methods described herein, the molecular tags do not comprise nucleotides that are altered by a procedure that affects a first nucleobase in the DNA differently from a second nucleobase in the DNA, such as any of those described herein (e.g., the tags do not comprise unmodified C where the procedure is bisulfite conversion or any other conversion that affects C; the tags do not comprise mC where the procedure is a conversion that affects mC; the tags do not comprise hmC where the procedure is a conversion that affects hmC; etc.).
Additional features of certain disclosed methods
A. Samples
[00382] A sample can be any biological sample isolated from a subject. A sample can be a bodily sample. Samples can include body tissues, such as known or suspected solid tumors, whole blood, platelets, serum, plasma, stool, red blood cells, white blood cells or leucocytes, endothelial cells, tissue biopsies, cerebrospinal fluid synovial fluid, lymphatic fluid, ascites fluid, interstitial or extracellular fluid, the fluid in spaces between cells, including gingival crevicular fluid, bone marrow, pleural effusions, cerebrospinal fluid, saliva, mucous, sputum, semen, sweat, urine. Samples are preferably body fluids, particularly blood and fractions thereof, and urine. A sample can be in the form originally isolated from a subject or can have been subjected to further processing to remove or add components, such as cells, or enrich for one component relative to another. Thus, a preferred body fluid for analysis is plasma or serum containing cell-free nucleic acids.
[00383] In some embodiments, a population of nucleic acids is obtained from a serum, plasma or blood sample from a subject suspected of having neoplasia, a tumor, precancer, or cancer or previously diagnosed with neoplasia, a tumor, precancer, or cancer. The population includes nucleic acids having varying levels of sequence variation, epigenetic variation, and/or post replication or transcriptional modifications. Postreplication modifications include modifications of cytosine, particularly at the 5-position of the nucleobase, e.g., 5-methylcytosine, 5- hydroxymethylcytosine, 5-formylcytosine and 5-carboxylcytosine.
[00384] A sample can be isolated or obtained from a subject and transported to a site of sample analysis. The sample may be preserved and shipped at a desirable temperature, e.g., room temperature, 4°C, -20°C, and/or -80°C. A sample can be isolated or obtained from a subject at the site of the sample analysis. The subject can be a human, a mammal, an animal, a companion animal, a service animal, or a pet. The subject may have a cancer, precancer, infection, transplant rejection, or other disease or disorder related to changes in the immune system. The subject may not have cancer or a detectable cancer symptom. The subject may have been treated with one or more cancer therapy, e.g., any one or more of chemotherapies, antibodies, vaccines or biologies. The subject may be in remission. The subject may or may not be diagnosed of being susceptible to cancer or any cancer-associated genetic mutations/disorders.
[00385] In some embodiments, the sample comprises plasma. The volume of plasma obtained can depend on the desired read depth for sequenced regions. Exemplary volumes are 0.4-40 ml, 5-20 ml, 10-20 ml. For examples, the volume can be 0.5 mL, 1 mL, 5 mL 10 mL, 20 mL, 30 mL, or 40 mL. A volume of sampled plasma may be 5 to 20 mL.
[00386] A sample can comprise various amount of nucleic acid that contains genome equivalents. For example, a sample of about 30 ng DNA can contain about 10,000 (104) haploid human genome equivalents and, in the case of cfDNA, about 200 billion (2xlOn) individual polynucleotide molecules. Similarly, a sample of about 100 ng of DNA can contain about 30,000 haploid human genome equivalents and, in the case of cfDNA, about 600 billion individual molecules.
[00387] A sample can comprise nucleic acids from different sources, e.g., from cells and cell-free of the same subject, from cells and cell-free of different subjects. A sample can comprise nucleic acids carrying mutations. For example, a sample can comprise DNA carrying germline mutations and/or somatic mutations. Germline mutations refer to mutations existing in germline DNA of a subject. Somatic mutations refer to mutations originating in somatic cells of a subject, e.g., precancer cells or cancer cells. A sample can comprise DNA carrying cancer-associated mutations (e.g., cancer-associated somatic mutations). A sample can comprise an epigenetic variant (i.e., a chemical or protein modification), wherein the epigenetic variant associated with the presence of a genetic variant such as a cancer-associated mutation. In some embodiments, the sample comprises an epigenetic variant associated with the presence of a genetic variant, wherein the sample does not comprise the genetic variant.
[00388] Exemplary amounts of cell-free nucleic acids in a sample before amplification range from about 1 fg to about 1 pg, e.g., 1 pg to 200 ng, 1 ng to 100 ng, 10 ng to 1000 ng. For example, the amount can be up to about 600 ng, up to about 500 ng, up to about 400 ng, up to about 300 ng, up to about 200 ng, up to about 100 ng, up to about 50 ng, or up to about 20 ng of cell-free nucleic acid molecules. The amount can be at least 1 fg, at least 10 fg, at least 100 fg, at least 1 pg, at least 10 pg, at least 100 pg, at least 1 ng, at least 10 ng, at least 100 ng, at least 150 ng, or at least 200 ng of cell-free nucleic acid molecules. The amount can be up to 1 femtogram (fg), 10 fg, 100 fg, 1 picogram (pg), 10 pg, 100 pg, 1 ng, 10 ng, 100 ng, 150 ng, or 200 ng of cell-free nucleic acid molecules. The method can comprise obtaining 1 femtogram (fg) to 200 ng- [0326] Cell-free nucleic acids are nucleic acids not contained within or otherwise bound to a cell or in other words nucleic acids remaining in a sample after removing intact cells. Cell- free nucleic acids include DNA, RNA, and hybrids thereof, including genomic DNA, mitochondrial DNA, siRNA, miRNA, circulating RNA (cRNA), tRNA, rRNA, small nucleolar RNA (snoRNA), Piwi-interacting RNA (piRNA), long non-coding RNA (long ncRNA), or fragments of any of these. Cell-free nucleic acids can be double-stranded, singlestranded, or a hybrid thereof. A cell-free nucleic acid can be released into bodily fluid through secretion or cell death processes, e.g., cellular necrosis and apoptosis. Some cell-free nucleic acids are released into bodily fluid from cancer cells e.g., circulating tumor DNA, (ctDNA). Others are released from healthy cells. In some embodiments, cfDNA is cell-free fetal DNA (cffDNA) In some embodiments, cell free nucleic acids are produced by tumor cells. In some embodiments, cell free nucleic acids are produced by a mixture of tumor cells and non-tumor cells.
[00389] Cell-free nucleic acids have an exemplary size distribution of about 100-500 nucleotides, with molecules of 110 to about 230 nucleotides representing about 90% of molecules, with a mode of about 168 nucleotides and a second minor peak in a range between 240 to 440 nucleotides.
[00390] Cell-free nucleic acids can be isolated from bodily fluids through a fractionation step in which cell-free nucleic acids, as found in solution, are separated from intact cells and other non-soluble components of the bodily fluid. Partitioning may include techniques such as centrifugation or filtration. Alternatively, cells in bodily fluids can be lysed and cell-free and cellular nucleic acids processed together. Generally, after addition of buffers and wash steps, nucleic acids can be precipitated with an alcohol. Further clean up steps may be used such as silica-based columns to remove contaminants or salts. Non-specific bulk carrier nucleic acids, such as C 1 DNA, DNA or protein for bisulfite sequencing, hybridization, and/or ligation, may be added throughout the reaction to optimize certain aspects of the procedure such as yield. [0329] After such processing, samples can include various forms of nucleic acid including double stranded DNA, single stranded DNA, and single stranded RNA. In some embodiments, single stranded DNA and RNA can be converted to double stranded forms so they are included in subsequent processing and analysis steps. [0330] DNA molecules can be linked to adapters at either one end or both ends. Typically, double-stranded molecules are blunt ended by treatment with a polymerase with a 5'-3' polymerase and a 3 '-5' exonuclease (or proof-reading function), in the presence of all four standard nucleotides. Klenow large fragment and T4 polymerase are examples of suitable polymerase. The blunt ended DNA molecules can be ligated with at least partially double stranded adapter (e.g., a Y shaped or bell-shaped adapter). Alternatively, complementary nucleotides can be added to blunt ends of sample nucleic acids and adapters to facilitate ligation. Contemplated herein are both blunt end ligation and sticky end ligation. In blunt end ligation, both the nucleic acid molecules and the adapter tags have blunt ends. In sticky-end ligation, typically, the nucleic acid molecules bear an “A” overhang and the adapters bear a “T” overhang.
B. Tags
[00391] Tags comprising barcodes can be incorporated into or otherwise joined to adapters. Tags can be incorporated by ligation, overlap extension PCR among other methods. i) Molecular tagging strategies
[00392] Molecular tagging refers to a tagging practice that allows one to differentiate among DNA molecules from which sequence reads originated. Tagging strategies can be divided into unique tagging and non-unique tagging strategies. In unique tagging, all or substantially all of the molecules in a sample bear a different tag, so that reads can be assigned to original molecules based on tag information alone. Tags used in such methods are sometimes referred to as “unique tags”. In non-unique tagging, different molecules in the same sample can bear the same tag, so that other information in addition to tag information is used to assign a sequence read to an original molecule. Such information may include start and stop coordinate, coordinate to which the molecule maps, start or stop coordinate alone, etc. Tags used in such methods are sometimes referred to as “non-unique tags”. Accordingly, it is not necessary to uniquely tag every molecule in a sample. It suffices to uniquely tag molecules falling within an identifiable class within a sample. Thus, molecules in different identifiable families can bear the same tag without loss of information about the identity of the tagged molecule.
[00393] In certain embodiments of non-unique tagging, the number of different tags used can be sufficient that there is a very high likelihood (e.g., at least 99%, at least 99.9%, at least 99.99% or at least 99.999% that all DNA molecules of a particular group bear a different tag. It is to be noted that when barcodes are used as tags, and when barcodes are attached, e.g., randomly, to both ends of a molecule, the combination of barcodes, together, can constitute a tag. This number, in term, is a function of the number of molecules falling into the calls. For example, the class may be all molecules mapping to the same start-stop position on a reference genome. The class may be all molecules mapping across a particular genetic locus, e.g., a particular base or a particular region (e.g., up to 100 bases or a gene or an exon of a gene). In certain embodiments, the number of different tags used to uniquely identify a number of molecules, z, in a class can be between any of 2*z, 3*z, 4*z, 5*z, 6*z, 7*z, 8*z, 9*z, 10*z, 11 *z, 12*z, 13*z, 14*z, 15*z, 16*z, 17*z, 18*z, 19*z, 20*z or 100*z (e.g., lower limit) and any of 100,000*z, 10,000*z, 1000*z or 100*z (e.g., upper limit).
[00394] For example, in a sample of about 5 ng to 30 ng of cell free DNA, one expects around 3000 molecules to map to a particular nucleotide coordinate, and between about 3 and 10 molecules having any start coordinate to share the same stop coordinate. Accordingly, about 50 to about 50,000 different tags (e.g., between about 6 and 220 barcode combinations) can suffice to uniquely tag all such molecules. To uniquely tag all 3000 molecules mapping across a nucleotide coordinate, about 1 million to about 20 million different tags would be required. [0336] Generally, assignment of unique or non-unique tags barcodes in reactions follows methods and systems described by US patent applications 20010053519, 20030152490, 20110160078, and U.S. Pat. No. 6,582,908 and U.S. Pat. No. 7,537,898 and US Pat. No. 9,598,731. Tags can be linked to sample nucleic acids randomly or non-randomly. [0337] The unique tags may be loaded so that more than about 1 , 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 50, 100, 500, 1000, 5000, 10000, 50,000, 100,000, 500,000, 1 ,000,000, 10,000,000, 50,000,000 or 1 ,000,000,000 unique tags are loaded per genome sample. In some cases, the unique tags may be loaded so that less than about 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 50, 100, 500, 1000, 5000, 10000, 50,000, 100,000, 500,000, 1 ,000,000, 10,000,000, 50,000,000 or 1 ,000,000,000 unique tags are loaded per genome sample. In some cases, the average number of unique tags loaded per sample genome is less than, or greater than, about 1 , 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 50, 100, 500, 1000, 5000, 10000, 50,000, 100,000, 500,000, 1 ,000,000, 10,000,000, 50,000,000 or 1 ,000,000,000 unique tags per genome sample.
[00395] A preferred format uses 20-50 different tags (e.g., barcodes) ligated to both ends of target nucleic acids. For example, 35 different tags (e.g., barcodes) ligated to both ends of target molecules creating 35 x 35 permutations, which equals 1225 for 35 tags. Such numbers of tags are sufficient so that different molecules having the same start and stop points have a high probability (e.g., at least 94%, 99.5%, 99.99%, 99.999%) of receiving different combinations of tags. Other barcode combinations include any number between 10 and 500, e.g., about 15x15, about 35x35, about 75x75, about 100x100, about 250x250, about 500x500.
[00396] In some cases, unique tags may be predetermined or random or semirandom sequence oligonucleotides. In other cases, a plurality of barcodes may be used such that barcodes are not necessarily unique to one another in the plurality. In this example, barcodes may be ligated to individual molecules such that the combination of the barcode and the sequence it may be ligated to creates a unique sequence that may be individually tracked. As described herein, detection of non-unique barcodes in combination with sequence data of beginning (start) and end (stop) portions of sequence reads may allow assignment of a unique identity to a particular molecule. The length or number of base pairs, of an individual sequence read may also be used to assign a unique identity to such a molecule. As described herein, fragments from a single strand of nucleic acid having been assigned a unique identity, may thereby permit subsequent identification of fragments from the parent strand. C. Amplification
[00397] Sample nucleic acids flanked by adapters can be amplified by PCR and other amplification methods. Amplification is typically primed by primers that anneal or bind to primer binding sites in adapters flanking a DNA molecule to be amplified. Amplification methods can involve cycles of denaturation, annealing and extension, resulting from thermocycling or can be isothermal as in transcription-mediated amplification. Other amplification methods include the ligase chain reaction, strand displacement amplification, nucleic acid sequence-based amplification, and selfsustained sequence-based replication.
[00398] In some embodiments, the present methods perform dsDNA ligations with T-tailed and C-tailed adapters, which result in amplification of at least 50, 60, 70 or 80% of double stranded nucleic acids before linking to adapters. Preferably the present methods increase the amount or number of amplified molecules relative to control methods performed with T-tailed adapters alone by at least 10, 15 or 20%.
D. Capture moieties.
[00399] As discussed above, nucleic acids in a sample can be subject to a capture step, in which molecules having target regions are captured for subsequent analysis. Target capture can involve use of probes (e.g., oligonucleotides) labeled with a capture moiety, such as biotin, and a second moiety or binding partner that binds to the capture moiety, such as streptavidin. In some embodiments, a capture moiety and binding partner can have higher and lower capture yields for different sets of target regions, such as those of the sequence-variable target region set and the epigenetic target region set, respectively, as discussed elsewhere herein. Methods comprising capture moieties are further described in, for example, U.S. patent 9,850,523, issuing December 26, 2017, which is incorporated herein by reference.
[00400] Capture moieties include, without limitation, biotin, avidin, streptavidin, a nucleic acid comprising a particular nucleotide sequence, a hapten recognized by an antibody, and magnetically attractable particles. The extraction moiety can be a member of a binding pair, such as biotin/ streptavidin or hapten/antibody. In some embodiments, a capture moiety that is attached to an analyte is captured by its binding pair which is attached to an isolatable moiety, such as a magnetically attractable particle or a large particle that can be sedimented through centrifugation. The capture moiety can be any type of molecule that allows affinity separation of nucleic acids bearing the capture moiety from nucleic acids lacking the capture moiety. Exemplary capture moieties are biotin which allows affinity separation by binding to streptavidin linked or linkable to a solid phase or an oligonucleotide, which allows affinity separation through binding to a complementary oligonucleotide linked or linkable to a solid phase.
E. Collections of target-specific probes
[00401] In some embodiments, a collection of target-specific probes is used in a method comprising an epigenetic target region set and/or a sequence-variable target region set, as described herein. In some embodiments, the collection of target-specific probes comprises target binding probes specific for a sequence-variable target region set and target-binding probes specific for an epigenetic target region set. In some embodiments, the capture yield of the target binding probes specific for the sequencevariable target region set is higher (e.g., at least 2-fold higher) than the capture yield of the target-binding probes specific for the epigenetic target region set. In some embodiments, the collection of target-specific probes is configured to have a capture yield specific for the sequence-variable target region set higher (e.g., at least 2-fold higher) than its capture yield specific for the epigenetic target region set.
[00402] In some embodiments, the capture yield of the target-binding probes specific for the sequence-variable target region set is at least 1.25-, 1.5-, 1.75-, 2-, 2.25- , 2.5-, 2.75-, 3-, 3.5-, 4-, 4.5-, 5-, 6-, 7- 8-, 9-, 10- 11 -, 12-, 13-, 14-, or 15-fold higher than the capture yield of the target-binding probes specific for the epigenetic target region set. In some embodiments, the capture yield of the target-binding probes specific for the sequence-variable target region set is 1 .25- to 1 .5-, 1.5- to 1 .75-, 1 .75- to 2-, 2- to 2.25-, 2.25- to 2.5-, 2.5- to 2.75-, 2.75- to 3-, 3- to 3.5-, 3.5- to 4-, 4- to 4.5- 4.5- to 5-, 5- to 5.5- , 5.5- to 6-, 6- to 7-, 7- to 8-, 8- to 9-, 9- to 10-, 10- to 11 -, 11 - to 12-, 13- to 14-, or 14- to 15-fold higher than the capture yield of the target-binding probes specific for the epigenetic target region set. [00403] In some embodiments, the collection of target-specific probes is configured to have a capture yield specific for the sequence-variable target region set at least 1 .25-, 1.5-, 1.75-, 2-, 2.25-, 2.5-, 2.75-, 3-, 3.5-, 4-, 4.5-, 5-, 6-, 7-, 8-, 9-, 10- 11 -, 12-, 13-, 14- or 15-fold higher than its capture yield for the epigenetic target region set. In some embodiments, the collection of target-specific probes is configured to have a capture yield specific for the sequence-variable target region set is 1 .25- to 1 .5-, 1.5- to 1 .75-, 1 .75- to 2-, 2- to 2.25-, 2.25- to 2.5-, 2.5- to 2.75-, 2.75- to 3-, 3- to 3.5-, 3.5- to 4-, 4- to 4.5-, 4.5- to 5-, 5- to 5.5-, 5.5- to 6-, 6- to 7-, 7- to 8-, 8- to 9-, 9- to 10-, 10- to 11 -, 11 - to 12-, 13- to 14-, or 14- to 15-fold higher than its capture yield specific for the epigenetic target region set.
[00404] The collection of probes can be configured to provide higher capture yields for the sequence-variable target region set in various ways, including concentration, different lengths and/or chemistries (e.g., that affect affinity), and combinations thereof. Affinity can be modulated by adjusting probe length and/or including nucleotide modifications as discussed below.
[00405] In some embodiments, the target-specific probes specific for the sequencevariable target region set are present at a higher concentration than the target-specific probes specific for the epigenetic target region set. In some embodiments, concentration of the target-binding probes specific for the sequence-variable target region set is at least 1.25-, 1.5-, 1.75- 2-, 2.25-, 2.5-, 2.75-, 3-, 3.5-, 4-, 4.5-, 5-, 6-, 7-, 8-, 9-, 10-, 11 - 12-, 13- , 14-, or 15-fold higher than the concentration of the target-binding probes specific for the epigenetic target region set. In some embodiments, the concentration of the targetbinding probes specific for the sequence-variable target region set is 1 .25- to 1 .5-, 1.5- to 1 .75-, 1 .75- to 2-, 2- to 2.25-, 2.25- to 2.5-, 2.5- to 2.75-, 2.75- to 3-, 3- to 3.5-, 3.5- to 4- 4- to 4.5-, 4.5- to 5-, 5- to 5.5-, 5.5- to 6-, 6- to 7-, 7- to 8-, 8- to 9-, 9- to 10-, 10- to I l l i - to 12-, 13- to 14-, or 14- to 15-fold higher than the concentration of the target-binding probes specific for the epigenetic target region set. In such embodiments, concentration may refer to the average mass per volume concentration of individual probes in each set. [00406] In some embodiments, the target-specific probes specific for the sequencevariable target region set have a higher affinity for their targets than the target-specific probes specific for the epigenetic target region set. Affinity can be modulated in any way known to those skilled in the art, including by using different probe chemistries. For example, certain nucleotide modifications, such as cytosine 5-methylation (in certain sequence contexts), modifications that provide a heteroatom at the T sugar position, and LNA nucleotides, can increase stability of double-stranded nucleic acids, indicating that oligonucleotides with such modifications have relatively higher affinity for their complementary sequences. See, e.g., Severin et ah, Nucleic Acids Res. 39: 8740-8751 (2011 ); Freier et ah, Nucleic Acids Res. 25: 4429-4443 (1997); US Patent No. 9,738,894. Also, longer sequence lengths will generally provide increased affinity. Other nucleotide modifications, such as the substitution of the nucleobase hypoxanthine for guanine, reduce affinity by reducing the amount of hydrogen bonding between the oligonucleotide and its complementary sequence. In some embodiments, the target-specific probes specific for the sequence-variable target region set have modifications that increase their affinity for their targets. In some embodiments, alternatively or additionally, the targetspecific probes specific for the epigenetic target region set have modifications that decrease their affinity for their targets. In some embodiments, the target-specific probes specific for the sequence- variable target region set have longer average lengths and/or higher average melting temperatures than the target-specific probes specific for the epigenetic target region set. These embodiments may be combined with each other and/or with differences in concentration as discussed above to achieve a desired fold difference in capture yield, such as any fold difference or range thereof described above. [00407] In some embodiments, the target-specific probes comprise a capture moiety. The capture moiety may be any of the capture moieties described herein, e.g., biotin. In some embodiments, the target-specific probes are linked to a solid support, e.g., covalently or non-covalently such as through the interaction of a binding pair of capture moieties. In some embodiments, the solid support is a bead, such as a magnetic bead.
[00408] In some embodiments, the target-specific probes specific for the sequencevariable target region set and/or the target-specific probes specific for the epigenetic target region set comprise a capture moiety as discussed above, e.g., probes comprising capture moieties and sequences selected to tile across a panel of regions, such as genes. [00409] In some embodiments, the target-specific probes are provided in a single composition. [00410] The single composition may be a solution (liquid or frozen). Alternatively, it may be a lyophilizate.
[00411] Alternatively, the target-specific probes may be provided as a plurality of compositions, e.g., comprising a first composition comprising probes specific for the epigenetic target region set and a second composition comprising probes specific for the sequence-variable target region set. These probes may be mixed in appropriate proportions to provide a combined probe composition with any of the foregoing fold differences in concentration and/or capture yield. Alternatively, they may be used in separate capture procedures (e.g., with aliquots of a sample or sequentially with the same sample) to provide first and second compositions comprising captured epigenetic target regions and sequence-variable target regions, respectively. i) Probes specific for epigenetic target regions.
[00412] The probes for the epigenetic target region set may comprise probes specific for one or more types of target regions likely to differentiate DNA originating from different types of immune cells, including rare immune cell types, and/or to differentiate DNA from precancerous or neoplastic (e.g., tumor or cancer) cells from healthy cells, e.g., non-neoplastic circulating cells. Exemplary types of such regions are discussed in detail herein. The probes for the epigenetic target region set may also comprise probes for one or more control regions, e.g., as described herein.
[00413] In some embodiments, the probes for the epigenetic target region probe set have a footprint of at least 100 kb, e.g., at least 200 kb, at least 300 kb, or at least 400 kb. In some embodiments, the probes for the epigenetic target region set have a footprint in the range of 100- 1000 kb, e.g., 100-200 kb, 200-300 kb, 300-400 kb, 400-500 kb, 500- 600 kb, 600-700 kb, 700- 800 kb, 800-900 kb, and 900-1 ,000 kb. In some embodiments, the probes for the epigenetic target region probe set have a footprint of at least 5 kb, e.g., at least 10, 20, or 50 kb. a. Hypermethylation target regions.
[00414] In some embodiments, for the methods using methylation-sensitive conversion (e.g., bisulfite or EM-seq), the probes can be designed to target either the converted molecules or unconverted molecules depending on the type of methylationsensitive conversion and the target region being enriched. For example, if bisulfite treatment is used, the unmethylated cytosines in the DNA molecules will be converted to dihydrouracil and methylated cytosines will remain unconverted as cytosine. For capturing DNA molecules in the hypermethylated target regions (where the molecules of interest to cancer or any other disease under investigation will be hypermethylated), the probes can be designed to capture the unconverted molecules, whereas for capturing molecules in the hypomethylated target regions (where the molecules of interest to cancer or any other disease under investigation will be hypomethylated or unmethylated), the probes can be designed to capture the converted molecules
[00415] In some embodiments, the probes for the epigenetic target region set comprise probes specific for one or more hypermethylation target regions. The hypermethylation target regions may be any of those set forth above. For example, in some embodiments, the probes specific for hypermethylation target regions comprise probes specific for a plurality of loci that are differentially methylated in different immune cell types. In some embodiments, each immune cell type specific hypermethylation target region comprises at least one CpG site that is methylated with a frequency greater than or equal to 0.3, 0.4, 0.5, or 0.6 in one immune cell type and with a frequency less than or equal to 0.1 , 0.2, or 0.3 in all other immune cell types. In some embodiments, each immune cell type specific hypermethylation target region comprises at least two CpG sites within 100 base pairs of each other that are each methylated with a frequency greater than or equal to 0.3, 0.4, 0.5, or 0.6 in one immune cell type and with a frequency less than or equal to 0.1 , 0.2, or 0.3 in all other immune cell types. In some such embodiments, each immune cell type specific hypermethylation target region comprises a total of at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 CpG sites within 150 base pairs or within 200 base pairs, wherein fewer than three of the at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 CpG sites are methylated with a frequency greater than 0.1 , 0.2, or 0.3 in any normal tissue type. In some embodiments, each immune cell type specific epigenetic target region set comprises at least 3, at least 5, at least 10, at least 20, or at least 30 hypermethylation target regions that are uniquely hypermethylated in each one of the immune cell types that are identified in the method.
[00416] In some embodiments, the probes specific for hypermethylation target regions comprise probes specific for a plurality of loci listed in Table 1 , e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the loci listed in Table 1. In some embodiments, the probes specific for hypermethylation target regions comprise probes specific for a plurality of loci listed in Table 2, e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the loci listed in Table 2. In some embodiments, the probes specific for hypermethylation target regions comprise probes specific for a plurality of loci listed in Table 1 or Table 2, e.g., at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, or 100% of the loci listed in Table 1 or Table 2.
[00417] In some embodiments, for each locus included as a target region, there may be one or more probes with a hybridization site that binds between the transcription start site and the stop codon (the last stop codon for genes that are alternatively spliced) of the gene. In some embodiments, the one or more probes bind within 300 bp of the listed position, e.g., within 200 or 100 bp. In some embodiments, a probe has a hybridization site overlapping the position listed above. In some embodiments, the probes specific for the hypermethylation target regions include probes specific for one, two, three, four, or five subsets of hypermethylation target regions that collectively show hypermethylation in one, two, three, four, or five of breast, colon, kidney, liver, and lung cancers, b. Hypomethylation target regions.
[00418] In some embodiments, the probes for the epigenetic target region set comprise probes specific for one or more hypomethylation target regions. The hypomethylation target regions may be any of those set forth above. For example, in some embodiments, the probes specific for hypomethylation target regions comprise probes specific for a plurality of loci that are differentially methylated in different immune cell types. In some embodiments, each immune cell type specific hypomethylation target region comprises at least one CpG site that is methylated with a frequency less than or equal to 0.1 , 0.2, or 0.3 in one immune cell type and with a frequency greater than or equal to 0.3, 0.4, 0.5, or 0.6 in all other immune cell types. In some embodiments, each immune cell type specific hypomethylation target region comprises at least two CpG sites within 100 base pairs of each other that are each methylated with a frequency less than or equal to 0.1 , 0.2, or 0.3 in one immune cell type and with a frequency greater than or equal to 0.3, 0.4, 0.5, or 0.6 in all other immune cell types. In some such embodiments, each immune cell type specific hypomethylation target region comprises a total of at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 CpG sites within 150 base pairs or within 200 base pairs, wherein fewer than three of the at least 2, 3, 4, 5, 6, 7, 8, 9, or 10 CpG sites are methylated with a frequency less than 0.1 , 0.2, or 0.3 in any normal tissue type. In some embodiments, each immune cell type specific epigenetic target region set comprises at least 3, at least 5, at least 10, at least 20, or at least 30 hypomethylation target regions that are uniquely hypomethylated in each one of the immune cell types that are identified in the method.
[00419] In some embodiments, the probes specific for one or more hypomethylation target regions may include probes for regions such as repeated elements, e.g., LINE1 elements, Alu elements, centromeric tandem repeats, pericentromeric tandem repeats, and satellite DNA, and intergenic regions that are ordinarily methylated in healthy cells may show reduced methylation in tumor cells.
[00420] In some embodiments, probes specific for hypomethylation target regions include probes specific for repeated elements and/or intergenic regions. In some embodiments, probes specific for repeated elements include probes specific for one, two, three, four, or five of LINE1 elements, Alu elements, centromeric tandem repeats, pericentromeric tandem repeats, and/or satellite DNA.
[00421] Exemplary probes specific for genomic regions that show cancer- associated hypomethylation include probes specific for nucleotides 8403565-8953708 and/or 151104701 - 151106035 of human chromosome 1. In some embodiments, the probes specific for hypomethylation target regions include probes specific for regions overlapping or comprising nucleotides 8403565-8953708 and/or 151104701 -151106035 of human chromosome 1 .
[00422] In some embodiments, the probes for the epigenetic target region set include probes specific for CTCF binding regions. In some embodiments, the probes specific for CTCF binding regions comprise probes specific for at least 10, 20, 50, 100, 200, or 500 CTCF binding regions, or 10-20, 20-50, 50-100, 100-200, 200-500, or SOO- WOO CTCF binding regions, e.g., such as CTCF binding regions described above or in one or more of CTCFBSDB or the Cuddapah et al., Martin et al., or Rhee et al. articles cited above. In some embodiments, the probes for the epigenetic target region set comprise at least 100 bp, at least 200 bp at least 300 bp, at least 400 bp, at least 500 bp, at least 750 bp, or at least 1000 bp upstream and downstream regions of the CTCF binding sites, d. Transcription start sites.
[00423] In some embodiments, the probes for the epigenetic target region set include probes specific for transcriptional start sites. In some embodiments, the probes specific for transcriptional start sites comprise probes specific for at least 10, 20, 50, 100, 200, or 500 transcriptional start sites, or 10-20, 20-50, 50-100, 100-200, 200-500, or 500- 1000 transcriptional start sites, e.g., such as transcriptional start sites listed in DBTSS. In some embodiments, the probes for the epigenetic target region set comprise probes for sequences at least 100 bp, at least 200 bp, at least 300 bp, at least 400 bp, at least 500 bp, at least 750 bp, or at least 1000 bp upstream and downstream of the transcriptional start sites.
[00424] As noted above, although focal amplifications are somatic mutations, they can be detected by sequencing based on read frequency in a manner analogous to approaches for detecting certain epigenetic changes such as changes in methylation. As such, regions that may show focal amplifications in cancer can be included in the epigenetic target region set, as discussed above. In some embodiments, the probes specific for the epigenetic target region set include probes specific for focal amplifications. In some embodiments, the probes specific for focal amplifications include probes specific for one or more of AR, BRAF, CCND1 , CCND2, CCNE1 , CDK4, CDK6, EGFR, ERBB2, FGFR1 , FGFR2, KIT, KRAS, MET, MYC, PDGFRA, PIK3CA, and RAFI. For example, in some embodiments, the probes specific for focal amplifications include probes specific for one or more of at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 , 12, 13, 14, 15, 16, 17, or 18 of the foregoing targets.
Control regions
[00425] In some embodiments, the probes specific for the epigenetic target region set include probes specific for positive control regions that are expected to be methylated in essentially all samples. In some embodiments, the probes specific for the epigenetic target region set include probes specific for negative control regions that are expected to be hypomethylated or unmethylated in essentially all samples.
2) Probes specific for sequence-variable target regions. [00426] The probes for the sequence-variable target region set may comprise probes specific for a plurality of regions known to undergo somatic mutations in cancer. The probes may be specific for any sequence-variable target region set described herein. Exemplary sequence-variable target region sets are discussed in detail herein, e.g., in the sections above concerning captured sets. [0366] In some embodiments, the sequence-variable target region probe set has a footprint of at least 0.5 kb, e.g., at least 1 kb, at least 2 kb, at least 5 kb, at least 10 kb, at least 20 kb, at least 30 kb, or at least 40 kb. In some embodiments, the epigenetic target region probe set has a footprint in the range of 0.5-100 kb, e.g., 0.5-2 kb, 2-10 kb, 10-20 kb, 20-30 kb, 30-40 kb, 40-50 kb, 50- 60 kb, 60-70 kb, 70-80 kb, 80-90 kb, and 90-100 kb.
[00427] In some embodiments, probes specific for the sequence-variable target region set comprise probes specific for at least a portion of at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, or at 70 of the genes of Table 4. In some embodiments, probes specific for the sequence- variable target region set comprise probes specific for the at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, or 70 of the SNVs of Table 3. In some embodiments, probes specific for the sequence-variable target region set comprise probes specific for at least 1 , at least 2, at least 3, at least 4, at least 5, or 6 of the fusions of Table 3. In some embodiments, probes specific for the sequence-variable target region set comprise probes specific for at least a portion of at least 1 , at least 2, or 3 of the indels of Table 4. In some embodiments, probes specific for the sequencevariable target region set comprise probes specific for at least a portion of at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, at least 70, or 73 of the genes of Table 5. In some embodiments, probes specific for the sequence-variable target region set comprise probes specific for at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, at least 60, at least 65, at least 70, or 73 of the SNVs of Table 5. In some embodiments, probes specific for the sequence-variable target region set comprise probes specific for at least 1 , at least 2, at least 3, at least 4, at least 5, or 6 of the fusions of Table 5. In some embodiments, probes specific for the sequence-variable target region set comprise probes specific for at least a portion of at least 1 , at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11 , at least 12, at least 13, at least 14, at least 15, at least 16, at least 17, or 18 of the indels of Table 5. In some embodiments, probes specific for the sequence-variable target region set comprise probes specific for at least a portion of at least 1 , at least 2, at least 3, at least 4, at least 5, at least 6, at least 7, at least 8, at least 9, at least 10, at least 11 , at least 12, at least 13, at least 14, at least 15, at least 16, at least 17, at least 18, at least 19, or at least 20 of the genes of Table 6.
[00428] In some embodiments, the probes specific for the sequence-variable target region set comprise probes specific for target regions from at least 10, 20, 30, or 35 cancer-related genes, such as AKTI, ALK, BRAF, CCND1 , CDK2A, CTNNB1 , EGFR, ERBB2, ESR1 , FGFR1 , FGFR2, FGFR3, FOXL2, GAT A3, GNA11 , GNAQ, GNAS, HRAS, IDH1 , IDH2, KIT, KRAS, MED 12, MET, MYC, NFE2L2, NRAS, PDGFRA, PIK3CA, PPP2R1A, PTEN, RET, STK11 , TP53, and U2AF 1.
Precision Treatments
[00429] The precision diagnostics provided by the improved computer system 110 may result in precision treatment plans, which may be identified by the computer system 110 (and/or curated by health professionals). For example, one type of precision diagnostic and treatment may relate to genes in the homologous recombination repair (HRR) pathway.
[00430] Homologous recombination is a type of genetic recombination in which nucleotide sequences are exchanged between two similar or identical molecules of DNA. It is most widely used by cells to accurately repair harmful breaks that occur on both strands of DNA, known as double-strand breaks (DSB). HRR provides a mechanism for the error-free removal of damage present in DNA that has replicated (S and G2 phases), to eliminate chromosomal breaks before the cell division occurs. The primary model for how homologous recombination repairs double-strand breaks in DNA is homologous recombination repair pathway which mediates the double-strand break repair (DSBR) pathway and the synthesis-dependent strand annealing (SDSA) pathway. Germ line and somatic deficiencies in homologous recombination genes have been strongly linked to breast, ovarian and prostate cancers.
[00431] The number and types of variant nucleotides in a sample can provide an indication of the amenability of the subject providing the sample to treatment, i.e., therapeutic intervention. For example, various poly ADP ribose polymerase (PARP) inhibitors have been shown to stop the growth of tumors from breast, ovarian and prostate cancers caused by hereditary mutations in the BRCA1 or BRCA2 genes. Some of these therapeutic agents may inhibit base excision repair (BER), which may compensate for the deficiency of HRR.
[00432] On the other hand, certain BRCA and HRR wildtype patients may not achieve clinical benefit from treatment with a PARP inhibitor. Furthermore, not all ovarian cancer patients with a BRCA mutation will respond to a PARP inhibitor. Moreover, different types of mutations may indicate different therapies. For example, somatic heterozygous deletions in HRR genes may indicate a different therapy than somatic homozygous deletions. Thus, the state of genetic material may influence therapy. In one example, a PARP inhibitor may be administered to an individual harboring a somatic homozygous deletion in a HRR gene, but not to an individual harboring a wildtype allele or somatic heterozygous deletions in the HRR gene.
[00433] In some implementations, a subject having HRD as determined by any of the methods disclosed may be administered a targeted therapy. The targeted therapy may comprise a PARP inhibitor. Examples of PARP inhibitors that may be administered include one or more of: VELIPARIB, OLAPARIB, TALAZOPARIB, RUCAPARIB, NIRAPARIB, PAMIPARIB, CEP 9722 (Cephalon), E7016 (Eisai), E7449 (Eisai, a PARP 1/2 and tankyrase 1/2 inhibitor), or 3-Aminobenzamide. In some implementations, the targeted therapy may comprise at least one base excision repair (BER) inhibitor. For example, OLAPARIB may inhibit BER. In certain implementations, the targeted therapy may comprise combination of a PARP inhibitor and radiotherapy. In an implementation, the combination of a PARP inhibitor and radiotherapy would permit the PARP inhibitor to lead to formation of double strand breaks from the single-strand breaks generated by the radiotherapy in tumor tissue (e.g., tissue with BRCA1/BRCA2 mutations). This combination can provide more powerful therapy per radiation dose. Related Therapies
[00434] In certain embodiments, the methods disclosed herein relate to identifying and administering therapies, such as customized therapies, to patients or subjects based on the determination of the presence or absence or levels of epigenomic and/or genetic variation. In some embodiments, the patient or subject has a given disease, disorder or condition, e.g., any of the cancers or other conditions described elsewhere herein. Essentially any cancer therapy (e.g., surgical therapy, radiation therapy, chemotherapy, immunotherapy, and/or the like) may be included as part of these methods.
[00435] Typically, the disease under consideration is a type of cancer. Non-limiting examples of such cancers include biliary tract cancer, bladder cancer, transitional cell carcinoma, urothelial carcinoma, brain cancer, gliomas, astrocytomas, breast cancer, metaplastic carcinoma, cervical cancer, cervical squamous cell carcinoma, rectal cancer, colorectal carcinoma, colon cancer, hereditary nonpolyposis colorectal cancer, colorectal adenocarcinomas, gastrointestinal stromal tumors (GISTs), endometrial carcinoma, endometrial stromal sarcomas, esophageal cancer, esophageal squamous cell carcinoma, esophageal adenocarcinoma, ocular melanoma, uveal melanoma, gallbladder carcinomas, gallbladder adenocarcinoma, renal cell carcinoma, clear cell renal cell carcinoma, transitional cell carcinoma, urothelial carcinomas, Wilms tumor, leukemia, acute lymphocytic leukemia (ALL), acute myeloid leukemia (AML), chronic lymphocytic leukemia (CLL), chronic myeloid leukemia (CML), chronic myelomonocytic leukemia (CMML), liver cancer, liver carcinoma, hepatoma, hepatocellular carcinoma, cholangiocarcinoma, hepatoblastoma, Lung cancer, non-small cell lung cancer (NSCLC), mesothelioma, B-cell lymphomas, non-Hodgkin lymphoma, diffuse large B-cell lymphoma, Mantle cell lymphoma, T cell lymphomas, non- Hodgkin lymphoma, precursor T-lymphoblastic lymphoma/leukemia, peripheral T cell lymphomas, multiple myeloma, nasopharyngeal carcinoma (NPC), neuroblastoma, oropharyngeal cancer, oral cavity squamous cell carcinomas, osteosarcoma, ovarian carcinoma, pancreatic cancer, pancreatic ductal adenocarcinoma, pseudopapillary neoplasms, acinar cell carcinomas, Prostate cancer, prostate adenocarcinoma, skin cancer, melanoma, malignant melanoma, cutaneous melanoma, small intestine carcinomas, stomach cancer, gastric carcinoma, gastrointestinal stromal tumor (GIST), uterine cancer, or uterine sarcoma.
[00436] Non-limiting examples of other genetic-based diseases, disorders, or conditions that are optionally evaluated using the methods and systems disclosed herein include achondroplasia, alpha- 1 antitrypsin deficiency, antiphospholipid syndrome, autism, autosomal dominant polycystic kidney disease, Charcot-Marie-Tooth (CMT), ch du chat, Crohn's disease, cystic fibrosis, Dercum disease, down syndrome, Duane syndrome, Duchenne muscular dystrophy, Factor V Leiden thrombophilia, familial hypercholesterolemia, familial mediterranean fever, fragile X syndrome, Gaucher disease, hemochromatosis, hemophilia, holoprosencephaly, Huntington's disease, Klinefelter syndrome, Marfan syndrome, myotonic dystrophy, neurofibromatosis, Noonan syndrome, osteogenesis imperfecta, Parkinson's disease, phenylketonuria, Poland anomaly, porphyria, progeria, retinitis pigmentosa, severe combined immunodeficiency (scid), sickle cell disease, spinal muscular atrophy, Tay-Sachs, thalassemia, trimethylaminuria, Turner syndrome, velocardiofacial syndrome, WAGR syndrome, Wilson disease, or the like.
[00437] In certain embodiments, the therapies can include one or more of treatments for target therapies, including abemaciclib (Verzenio), abiraterone acetate (Zytiga), acalabrutinib (Calquence), adagrasib (Krazati), ado-trastuzumab emtansine (Kadcyla), afatinib dimaleate (Gilotrif), alectinib (Alecensa), alemtuzumab (Campath), alitretinoin (Panretin), alpelisib (Piqray), amivantamab- vmjw (Rybrevant), anastrozole (Arimidex), apalutamide (Erleada), asciminib hydrochloride (Scemblix), atezolizumab (Tecentriq), atezolizumab (Tecentriq), avapritinib (Ayvakit), avelumab (Bavencio), axicabtagene ciloleucel (Yescarta), axitinib (Inlyta), belinostat (Beleodaq), belzutifan (Welireg), bevacizumab (Avastin), bexarotene (Targretin), binimetinib (Mektovi), blinatumomab (Blincyto), bortezomib (Velcade), bosutinib (Bosulif), brentuximab vedotin (Adcetris), brexucabtagene autoleucel (Tecartus), brigatinib (Alunbrig), cabazitaxel (Jevtana), cabozantinib-s-malate (Cabometyx), cabozantinib-s-malate (Cometriq), capmatinib hydrochloride (Tabrecta), carfilzomib (Kyprolis), cemiplimab-rwlc (Libtayo), ceritinib (Zykadia), cetuximab (Erbitux), ciltacabtagene autoleucel (Carvykti), cobimetinib fumarate (Cotellic), copanlisib hydrochloride (Aliqopa), crizotinib (Xalkori), dabrafenib (Tafmlar), dabrafenib mesylate (Tafmlar), dacomitinib (Vizimpro), daratumumab (Darzalex), daratumumab and hyaluronidase-fihj (Darzalex Faspro), darolutamide (Nubeqa), dasatinib (Sprycel), denileukin diftitox (Ontak), denosumab (Xgeva), dinutuximab (Unituxin), dostarlimab-gxly (Jemperli), durvalumab (Imfinzi), duvelisib (Copiktra), elacestrant dihydrochloride (Orserdu), elotuzumab (Empliciti), enasidenib mesylate (Idhifa), encorafenib (Braftovi), enfortumab vedotin-ejfv (Padcev), entrectinib (Rozlytrek), enzalutamide (Xtandi), erdafitinib (Balversa), erlotinib hydrochloride (Tarceva), everolimus (Afinitor), exemestane (Aromasin), fam -trastuzumab deruxtecan- nxki (Enhertu), fam-trastuzumab deruxtecan-nxki (Enhertu), fedratinib hydrochloride (Inrebic), fulvestrant (Faslodex), futibatinib (Lytgobi), gefitinib (Iressa), gemtuzumab ozogamicin (Mylotarg), gilteritinib fumarate (Xospata), glasdegib maleate (Daurismo), ibritumomab tiuxetan (Zevalin), ibrutinib (Imbruvica), idecabtagene vicleucel (Abecma), idelalisib (Zydelig), imatinib mesylate (Gleevec), infigratinib phosphate (Truseltiq), inotuzumab ozogamicin (Besponsa), iobenguane 1 131 (Azedra), ipilimumab (Yervoy), isatuximab-irfc (Sarclisa), ivosidenib (Tibsovo), ixazomib citrate (Ninlaro), lanreotide acetate (SomatulineDepot), lapatinib ditosylate (Tykerb), larotrectinib sulfate (Vitrakvi), lenvatinib mesylate (Lenvima), letrozole (Femara), lisocabtagene maraleucel (Breyanzi), loncastuximab tesirine-lpyl (Zynlonta), lorlatinib (Lorbrena), lutetium Lu 177 vipivotide tetraxetan (Pluvicto), lutetium Lu 177-dotatate (Lutathera), margetuximab-cmkb (Margenza), midostaurin (Rydapt), mirvetuximab soravtansine-gynx (Elahere), mobocertinib succinate (Exkivity), mogamulizumab-kpkc (Poteligeo), mosunetuzumab- axgb (Lunsumio), moxetumomab pasudotox-tdfk(Lumoxiti), naxitamab-gqgk (Danyelza), necitumumab (Portrazza), neratinib maleate (Nerlynx), nilotinib (Tasigna), niraparib tosylate monohydrate (Zejula), nivolumab (Opdivo), nivolumab and relatlimab-rmbw (Opdualag), obinutuzumab (Gazyva), ofatumumab (Arzerra), olaparib (Lynparza), olutasidenib (Rezlidhia), osimertinib mesylate (Tagrisso), pacritinib citrate (Vonjo), palbociclib (Ibrance), panitumumab (Vectibix), pazopanib hydrochloride(Votrient), pembrolizumab (Keytruda), pemigatinib(Pemazyre), pertuzumab (Perjeta), pertuzumab, trastuzumab, and hyaluronidase-zzxf (Phesgo), pexidartinib hydrochloride (Turalio), pirtobrutinib (Jaypirca), polatuzumab vedotin-piiq (Polivy), ponatinib hydrochloride (Iclusig), pralatrexate (Folotyn), pralsetinib (Gavreto), radium 223 dichloride (Xofigo), ramucirumab (Cyramza), regorafenib (Stivarga), retifanlimab-dlwr (Zynyz), ribociclib (Kisqali), ripretinib (Qinlock), rituximab (Rituxan), rituximab and hyaluronidase human (Rituxan Hycela), romidepsin (Istodax), rucaparib camsylate(Rubraca), ruxolitinib phosphate (Jakafi), sacituzumab govitecan-hziy (Trodelvy), selinexor (Xpovio), selpercatinib (Retevmo), selumetinib sulfate (Koselugo), siltuximab (Sylvant), sirolimus protein-bound particles (Fyarro), sonidegib (Odomzo), sorafenib tosylate (Nexavar), sotorasib (Lumakras), sunitinib malate (Sutent), tafasitamab-cxix (Monjuvi), tagraxofusp- erzs (Elzonris), talazoparib tosylate (Talzenna), tamoxifen citrate (Soltamox), tazemetostat hydrobromide (Tazverik), tebentafusp-tebn (Kimmtrak), teclistamab-cqyv (Tecvayli), temsirolimus (Torisel), tepotinib hydrochloride (Tepmetko), tisagenlecleucel (Kymriah), tisotumab vedotin-tftv (Tivdak), tivozanib hydrochloride (Fotivda), toremifene (Fareston), trametinib (Mekinist), trametinib dimethyl sulfoxide (Mekinist), trastuzumab (Herceptin), tremelimumab-actl (Imjudo), tretinoin (Vesanoid), tucatinib (Tukysa), vandetanib (Caprelsa), vemurafenib (Zelboraf), venetoclax (Venclexta), vismodegib (Erivedge), vorinostat (Zolinza), zanubrutinib (Brukinsa), ziv-aflibercept (Zaltrap).
[00438] In certain embodiments, the therapy administered to a subject comprises at least one chemotherapy drug. In some embodiments, the chemotherapy drug may comprise alkylating agents (for example, but not limited to, Chlorambucil, Cyclophosphamide, Cisplatin and Carboplatin), nitrosoureas (for example, but not limited to, Carmustine and Lomustine), anti-metabolites (for example, but not limited to, Fluorauracil, Methotrexate and Fludarabine), plant alkaloids and natural products (for example, but not limited to, Vincristine, Paclitaxel and Topotecan), anti- tumor antibiotics (for example, but not limited to, Bleomycin, Doxorubicin and Mitoxantrone), hormonal agents (for example, but not limited to, Prednisone, Dexamethasone, Tamoxifen and Leuprolide) and biological response modifiers (for example, but not limited to, Herceptin and Avastin, Erbitux and Rituxan). In some embodiments, the chemotherapy administered to a subject may comprise FOLFOX or FOLFIRI. In certain embodiments, a therapy may be administered to a subject that comprises at least one PARP inhibitor. In certain embodiments, the PARP inhibitor may include OLAPARIB, TALAZOPARIB, RUCAPARIB, NIRAPARIB (trade name ZEJULA), among others. In some embodiments, the methods comprise administering a therapy comprising a PARP inhibitor, such as olaparib, to a subject determined to have homologous recombination repair (HRR) gene or deficiency (HRD), such as with BRCA1 , BRCA2, ATM, BARD1 , BRIP1 , CDK12, CHEK1 , CHEK2, FANCL, PALB2, RAD51 B, RAD51 C, RAD51 D, and RAD54L alterations. In some embodiments, the subject has a metastatic castrate resistant prostate cancer (mCRPC). In some embodiments, the PARP inhibitor, such as olaprib is used to treat a subject having ovarian cancer, breast cancer, pancreatic cancer, or mCRPC, wherein the subject is determined to have alterations in BRCA1 , BRCA2, and/or ATM.
[00439] In some embodiments, essentially any cancer therapy (e.g., surgical therapy, radiation therapy, chemotherapy, immunotherapy, and/or the like) may be included as part of these methods. Customized therapies can include at least one immunotherapy (or an immunotherapeutic agent). Immunotherapy refers generally to methods of enhancing an immune response against a given cancer type. In certain embodiments, immunotherapy refers to methods of enhancing a T cell response against a tumor or cancer.
[00440] In some embodiments, the immunotherapy or immunotherapeutic agent targets an immune checkpoint molecule. Certain tumors are able to evade the immune system by co-opting an immune checkpoint pathway. Thus, targeting immune checkpoints has emerged as an effective approach for countering a tumor’s ability to evade the immune system and activating anti-tumor immunity against certain cancers. Pardoll, Nature Reviews Cancer, 2012, 12:252-264.
[00441] In certain embodiments, the immune checkpoint molecule is an inhibitory molecule that reduces a signal involved in the T cell response to antigen. For example, CTLA4 is expressed on T cells and plays a role in downregulating T cell activation by binding to CD80 (aka B7.1 ) or CD86 (aka B7.2) on antigen presenting cells. PD-1 is another inhibitory checkpoint molecule that is expressed on T cells. PD-1 limits the activity of T cells in peripheral tissues during an inflammatory response. In addition, the ligand for PD-1 (PD-L1 or PD-L2) is commonly upregulated on the surface of many different tumors, resulting in the downregulation of anti-tumor immune responses in the tumor microenvironment. In certain embodiments, the inhibitory immune checkpoint molecule is CTLA4 or PD-1. In other embodiments, the inhibitory immune checkpoint molecule is a ligand for PD-1 , such as PD-L1 or PD-L2. In other embodiments, the inhibitory immune checkpoint molecule is a ligand for CTLA4, such as CD80 or CD86. In other embodiments, the inhibitory immune checkpoint molecule is lymphocyte activation gene 3 (LAG3), killer cell immunoglobulin like receptor (KIR), T cell membrane protein 3 (TIM3), galectin 9 (GAL9), or adenosine A2a receptor (A2aR).
[00442] Antagonists that target these immune checkpoint molecules can be used to enhance antigen-specific T cell responses against certain cancers. Accordingly, in certain embodiments, the immunotherapy or immunotherapeutic agent is an antagonist of an inhibitory immune checkpoint molecule. In certain embodiments, the inhibitory immune checkpoint molecule is PD-1. In certain embodiments, the inhibitory immune checkpoint molecule is PD-L1. In certain embodiments, the antagonist of the inhibitory immune checkpoint molecule is an antibody (e.g., a monoclonal antibody). In certain embodiments, the antibody or monoclonal antibody is an anti-CTLA4, anti-PD-1 , anti-PD- L1 , or anti-PD-L2 antibody. In certain embodiments, the antibody is a monoclonal anti- PD-1 antibody. In some embodiments, the antibody is a monoclonal anti-PD-L1 antibody. In certain embodiments, the monoclonal antibody is a combination of an anti-CTLA4 antibody and an anti-PD-1 antibody, an anti-CTLA4 antibody and an anti-PD-L1 antibody, or an anti-PD-L1 antibody and an anti-PD-1 antibody. In certain embodiments, the anti- PD-1 antibody is one or more of pembrolizumab (Keytruda®) or nivolumab (Opdivo®). In certain embodiments, the anti-CTLA4 antibody is ipilimumab (Yervoy®). In certain embodiments, the anti-PD-L1 antibody is one or more of atezolizumab (Tecentriq®), avelumab (Bavencio®), or durvalumab (Imfinzi®). In certain embodiments, immunotherapy, such as pembrolizumab, is used to treat a subject determined to have a high microsatellite instability status (MSI-H). In certain embodiments, the immunotherapy, such as pembrolizumab, is used to treat a subject determined to have a high tumor mutational burden (TMB), for example, then the TMB status is greater than or equal to 10 mutations per megabase. In certain embodiment, the immunotherapy, such as pembrolizumab, is used to treat a subject determined to a have a mismatch repair deficiency (dMMR), such as in genes comprising MLH1 , PMS2, MSH2 and MSH6.
[00443] In certain embodiments, the immunotherapy or immunotherapeutic agent is an antagonist (e.g., antibody) against CD80, CD86, LAG3, KIR, TIM3, GAL9, or A2aR. In other embodiments, the antagonist is a soluble version of the inhibitory immune checkpoint molecule, such as a soluble fusion protein comprising the extracellular domain of the inhibitory immune checkpoint molecule and an Fc domain of an antibody. In certain embodiments, the soluble fusion protein comprises the extracellular domain of CTLA4, PD-1 , PD-L1 , or PD-L2. In some embodiments, the soluble fusion protein comprises the extracellular domain of CD80, CD86, LAG3, KIR, TIM3, GAL9, or A2aR. In one embodiment, the soluble fusion protein comprises the extracellular domain of PD-L2 or LAG3.
[00444] In certain embodiments, the immune checkpoint molecule is a costimulatory molecule that amplifies a signal involved in a T cell response to an antigen. For example, CD28 is a co-stimulatory receptor expressed on T cells. When a T cell binds to antigen through its T cell receptor, CD28 binds to CD80 (aka B7.1 ) or CD86 (aka B7.2) on antigen-presenting cells to amplify T cell receptor signaling and promote T cell activation. Because CD28 binds to the same ligands (CD80 and CD86) as CTLA4, CTLA4 is able to counteract or regulate the co-stimulatory signaling mediated by CD28. In certain embodiments, the immune checkpoint molecule is a co-stimulatory molecule selected from CD28, inducible T cell co-stimulator (ICOS), CD137, 0X40, or CD27. In other embodiments, the immune checkpoint molecule is a ligand of a co-stimulatory molecule, including, for example, CD80, CD86, B7RP1 , B7-H3, B7-H4, CD137L, OX40L, or CD70. [00445] Agonists that target these co-stimulatory checkpoint molecules can be used to enhance antigen-specific T cell responses against certain cancers. Accordingly, in certain embodiments, the immunotherapy or immunotherapeutic agent is an agonist of a co-stimulatory checkpoint molecule. In certain embodiments, the agonist of the co- stimulatory checkpoint molecule is an agonist antibody and preferably is a monoclonal antibody. In certain embodiments, the agonist antibody or monoclonal antibody is an anti- CD28 antibody. In other embodiments, the agonist antibody or monoclonal antibody is an anti-ICOS, anti-CD137, anti-OX40, or anti-CD27 antibody. In other embodiments, the agonist antibody or monoclonal antibody is an anti-CD80, anti-CD86, anti-B7RP1 , anti- B7-H3, anti-B7-H4, anti-CD137L, anti-OX40L, or anti-CD70 antibody.
[00446] In certain embodiments, the status of a nucleic acid variant from a sample from a subject as being of somatic or germline origin may be compared with a database of comparator results from a reference population to identify customized or targeted therapies for that subject. Typically, the reference population includes patients with the same cancer or disease type as the subject and/or patients who are receiving, or who have received, the same therapy as the subject. A customized or targeted therapy (or therapies) may be identified when the nucleic variant and the comparator results satisfy certain classification criteria (e.g., are a substantial or an approximate match).
[00447] In certain embodiments, the customized therapies described herein are typically administered parenterally (e.g., intravenously or subcutaneously). Pharmaceutical compositions containing an immunotherapeutic agent are typically administered intravenously. Certain therapeutic agents are administered orally. However, customized therapies (e.g., immunotherapeutic agents, etc.) may also be administered by any method known in the art, for example, buccal, sublingual, rectal, vaginal, intraurethral, topical, intraocular, intranasal, and/or intraauricular, which administration may include tablets, capsules, granules, aqueous suspensions, gels, sprays, suppositories, salves, ointments, or the like.
[00448] In certain embodiments, the present methods are also useful in determining the efficacy of particular treatment options. For example, the number of variations detected, irrespective of their precise identity, is a predictor of amenability to immunotherapy because the mutations create neoepitopes that can be subject of immune attack (see e.g., US20200370129).
[00449] Other variations or copy number variations indicate suitability of a particular drug. Some examples of such variations are as follows:
Table 7: List of cancer types with associated biomarker target and drug
[00450] In certain embodiments, the therapy comprises administrating a treatment to a subject determined to have a copy number amplification. In some embodiments, the treatment may comprise trastuzumab, ado-trastuzumab emtansine, or pertuzumab where the subject was determined to have an ERBB2 (HER2) gene amplification. In some embodiments, the subject has breast cancer or gastric cancer.
[00451] In some embodiments, the therapy comprises administering one or more drugs to the subject. For example, patients with non-small lung cancer determined to have either an EGFR exon 19 deletion or an EGFR exon 21 L858R alteration may be treated with amivantamab in combination with lazertinib.
[00452] The present methods can be used to generate or profile, fingerprint or set of data that is a summation of genetic information derived from different cells in a heterogeneous disease. This set of data may comprise copy number variation, nucleotide variation, epigenomic information, and/or tumor fraction. In some embodiments, the methods disclosed herein are used to monitor the efficacy or responsiveness of a treatment to the subject. In some embodiments, the methods disclosed herein can be used to determine whether the subject is a candidate for a therapy to treat the cancer or disease.
[00453] The present methods can be used to diagnose, prognose, monitor or observe cancers or other diseases of fetal origin. That is, these methodologies can be employed in a pregnant subject to diagnose, prognose, monitor or observe cancers or other diseases in an unborn subject whose DNA and other nucleic acids may co-circulate with maternal molecules.
[00454] In certain embodiments, the present methods can be used to determine minimal residual disease (MRD) of a subject, for example, based on a tumor fraction determination. In some embodiments, the methods may be directed to determining MRD by using a tissue-informed assay (i.e. , using a tissue sample collected from a patient to determine a personalized panel to enrich for one or more genomic and/or epigenomic variants in a subsequent blood sample from the patient) or a tissue-naTve assay.
[00455] In certain embodiments, the present methods can integrate genomic and/or epigenomic data with proteomic (proteins and their post-translational modifications), transcriptom ic, fragmentomic, immunological, histological, and/or other analyte-specific data to determine disease initiation, progression, malignant transformation, and therapeutic outcomes.
[00456] Figure 5 is a block diagram illustrating components of a machine 500, according to some example implementations, able to read instructions from a machine- readable medium (e.g., a machine-readable storage medium) and perform any one or more of the methodologies discussed herein. Specifically, Figure 5 shows a diagrammatic representation of the machine 500 in the example form of a computer system, within which instructions 502 (e.g., software, a program, an application, an applet, an app, or other executable code) for causing the machine 500 to perform any one or more of the methodologies discussed herein may be executed. As such, the instructions 502 may be used to implement modules or components described herein. The instructions 502 transform the general, non-programmed machine 500 into a particular machine 500 programmed to carry out the described and illustrated functions in the manner described. In alternative implementations, the machine 500 operates as a standalone device or may be coupled (e.g., networked) to other machines. In a networked deployment, the machine 500 may operate in the capacity of a server machine or a client machine in a server-client network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. The machine 500 may comprise, but not be limited to, a server computer, a client computer, a personal computer (PC), a tablet computer, a laptop computer, a netbook, a set-top box (STB), a personal digital assistant (PDA), an entertainment media system, a cellular telephone, a smart phone, a mobile device, a wearable device (e.g., a smart watch), a smart home device (e.g., a smart appliance), other smart devices, a web appliance, a network router, a network switch, a network bridge, or any machine capable of executing the instructions 502, sequentially or otherwise, that specify actions to be taken by machine 500. Further, while only a single machine 500 is illustrated, the term "machine" shall also be taken to include a collection of machines that individually or jointly execute the instructions 502 to perform any one or more of the methodologies discussed herein.
[00457] The machine 500 may include processors 504, memory/storage 506, and I/O components 508, which may be configured to communicate with each other such as via a bus 510. In an example implementation, the processors 504 (e.g., a central processing unit (CPU), a reduced instruction set computing (RISC) processor, a complex instruction set computing (CISC) processor, a graphics processing unit (GPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a radiofrequency integrated circuit (RFIC), another processor, or any suitable combination thereof) may include, for example, a processor 512 and a processor 514 that may execute the instructions 502. The term “processor” is intended to include multi-core processors 504 that may comprise two or more independent processors (sometimes referred to as “cores”) that may execute instructions 502 contemporaneously. Although Figure 5 shows multiple processors 504, the machine 500 may include a single processor 512 with a single core, a single processor 512 with multiple cores (e.g., a multi-core processor), multiple processors 512, 514 with a single core, multiple processors 512, 514 with multiple cores, or any combination thereof.
[00458] The memory/storage 506 may include memory, such as a main memory 516, or other memory storage, and a storage unit 518, both accessible to the processors 504 such as via the bus 510. The storage unit 518 and main memory 516 store the instructions 502 embodying any one or more of the methodologies or functions described herein. The instructions 502 may also reside, completely or partially, within the main memory 516, within the storage unit 518, within at least one of the processors 504 (e.g., within the processor’s cache memory), or any suitable combination thereof, during execution thereof by the machine 500. Accordingly, the main memory 516, the storage unit 518, and the memory of processors 504 are examples of machine-readable media.
[00459] The I/O components 508 may include a wide variety of components to receive input, provide output, produce output, transmit information, exchange information, capture measurements, and so on. The specific I/O components 508 that are included in a particular machine 500 will depend on the type of machine. For example, portable machines such as mobile phones will likely include a touch input device or other such input mechanisms, while a headless server machine will likely not include such a touch input device. It will be appreciated that the I/O components 508 may include many other components that are not shown in Figure 5. The I/O components 508 are grouped according to functionality merely for simplifying the following discussion and the grouping is in no way limiting. In various example implementations, the I/O components 508 may include user output components 520 and user input components 522. The user output components 520 may include visual components (e.g., a display such as a plasma display panel (PDP), a light emitting diode (LED) display, a liquid crystal display (LCD), a projector, or a cathode ray tube (CRT)), acoustic components (e.g., speakers), haptic components (e.g., a vibratory motor, resistance mechanisms), other signal generators, and so forth. The user input components 522 may include alphanumeric input components (e.g., a keyboard, a touch screen configured to receive alphanumeric input, a photo-optical keyboard, or other alphanumeric input components), point-based input components (e.g., a mouse, a touchpad, a trackball, a joystick, a motion sensor, or other pointing instrument), tactile input components (e.g., a physical button, a touch screen that provides location or force of touches or touch gestures, or other tactile input components), audio input components (e.g., a microphone), and the like.
[00460] In further example implementations, the I/O components 508components 508 may include biometric components 524, motion components 526, environmental components 528, or position components 530 among a wide array of other components. For example, the biometric components 524 may include components to detect expressions (e.g., hand expressions, facial expressions, vocal expressions, body gestures, or eye tracking), measure biosignals (e.g., blood pressure, heart rate, body temperature, perspiration, or brain waves), identify a person (e.g., voice identification, retinal identification, facial identification, fingerprint identification, or electroencephalogram -based identification), and the like. The motion components 526 may include acceleration sensor components (e.g., accelerometer), gravitation sensor components, rotation sensor components (e.g., gyroscope), and so forth. The environmental components 528 may include, for example, illumination sensor components (e.g., photometer), temperature sensor components (e.g., one or more thermometer that detect ambient temperature), humidity sensor components, pressure sensor components (e.g., barometer), acoustic sensor components (e.g., one or more microphones that detect background noise), proximity sensor components (e.g., infrared sensors that detect nearby objects), gas sensors (e.g., gas detection sensors to detection concentrations of hazardous gases for safety or to measure pollutants in the atmosphere), or other components that may provide indications, measurements, or signals corresponding to a surrounding physical environment. The position components 530 may include location sensor components (e.g., a GPS receiver component), altitude sensor components (e.g., altimeters or barometers that detect air pressure from which altitude may be derived), orientation sensor components (e.g., magnetometers), and the like.
[00461] Communication may be implemented using a wide variety of technologies. The I/O components 508 may include communication components 532 operable to couple the machine 500 to a network 534 or devices 536. For example, the communication components 532 may include a network interface component or other suitable device to interface with the network 534. In further examples, communication components 532 may include wired communication components, wireless communication components, cellular communication components, near field communication (NFC) components, Bluetooth® components (e.g., Bluetooth® Low Energy), Wi-Fi® components, and other communication components to provide communication via other modalities. The devices 536 may be another machine 500 or any of a wide variety of peripheral devices (e.g., a peripheral device coupled via a USB). [00462] Moreover, the communication components 532 may detect identifiers or include components operable to detect identifiers. For example, the communication components 532 may include radio frequency identification (RFID) tag reader components, NFC smart tag detection components, optical reader components (e.g., an optical sensor to detect one-dimensional bar codes such as Universal Product Code (UPC) bar code, multi-dimensional bar codes such as Quick Response (QR) code, Aztec code, Data Matrix, Dataglyph, MaxiCode, PDF417, Ultra Code, UCC RSS-2D bar code, and other optical codes), or acoustic detection components (e.g., microphones to identify tagged audio signals). In addition, a variety of information may be derived via the communication components 532, such as location via Internet Protocol (IP) geo-location, location via Wi-Fi® signal triangulation, location via detecting an NFC beacon signal that may indicate a particular location, and so forth.
[00463] As used herein, “component” refers to a device, physical entity, or logic having boundaries defined by function or subroutine calls, branch points, APIs, or other technologies that provide for the partitioning or modularization of particular processing or control functions. Components may be combined via their interfaces with other components to carry out a machine process. A component may be a packaged functional hardware unit designed for use with other components and a part of a program that usually performs a particular function of related functions. Components may constitute either software components (e.g., code embodied on a machine-readable medium) or hardware components. A "hardware component" is a tangible unit capable of performing certain operations and may be configured or arranged in a certain physical manner. In various example implementations, one or more computer systems (e.g., a standalone computer system, a client computer system, or a server computer system) or one or more hardware components of a computer system (e.g., a processor or a group of processors) may be configured by software (e.g., an application or application portion) as a hardware component that operates to perform certain operations as described herein.
[00464] A hardware component may also be implemented mechanically, electronically, or any suitable combination thereof. For example, a hardware component may include dedicated circuitry or logic that is permanently configured to perform certain operations. A hardware component may be a special-purpose processor, such as a field- programmable gate array (FPGA) or an ASIC. A hardware component may also include programmable logic or circuitry that is temporarily configured by software to perform certain operations. For example, a hardware component may include software executed by a general-purpose processor 504 or other programmable processor. Once configured by such software, hardware components become specific machines (or specific components of a machine 500) uniquely tailored to perform the configured functions and are no longer general-purpose processors 504. It will be appreciated that the decision to implement a hardware component mechanically, in dedicated and permanently configured circuitry, or in temporarily configured circuitry (e.g., configured by software) may be driven by cost and time considerations. Accordingly, the phrase "hardware component"(or "hardware-implemented component") should be understood to encompass a tangible entity, be that an entity that is physically constructed, permanently configured (e.g., hardwired), or temporarily configured (e.g., programmed) to operate in a certain manner or to perform certain operations described herein. Considering implementations in which hardware components are temporarily configured (e.g., programmed), each of the hardware components need not be configured or instantiated at any one instance in time. For example, where a hardware component comprises a general-purpose processor 504 configured by software to become a special-purpose processor, the general-purpose processor 504 may be configured as respectively different special-purpose processors (e.g., comprising different hardware components) at different times. Software accordingly configures a particular processor 512, 514 or processors 504, for example, to constitute a particular hardware component at one instance of time and to constitute a different hardware component at a different instance of time.
[00465] Hardware components can provide information to, and receive information from, other hardware components. Accordingly, the described hardware components may be regarded as being communicatively coupled. Where multiple hardware components exist contemporaneously, communications may be achieved through signal transmission (e.g., over appropriate circuits and buses) between or among two or more of the hardware components. In implementations in which multiple hardware components are configured or instantiated at different times, communications between such hardware components may be achieved, for example, through the storage and retrieval of information in memory structures to which the multiple hardware components have access. For example, one hardware component may perform an operation and store the output of that operation in a memory device to which it is communicatively coupled. A further hardware component may then, at a later time, access the memory device to retrieve and process the stored output.
[00466] Hardware components may also initiate communications with input or output devices, and can operate on a resource (e.g., a collection of information). The various operations of example methods described herein may be performed, at least partially, by one or more processors 504 that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors 504 may constitute processor- implemented components that operate to perform one or more operations or functions described herein. As used herein, "processor-implemented component" refers to a hardware component implemented using one or more processors 504. Similarly, the methods described herein may be at least partially processor-implemented, with a particular processor 512, 514 or processors 504 being an example of hardware. For example, at least some of the operations of a method may be performed by one or more processors 504 or processor-implemented components. Moreover, the one or more processors 504 may also operate to support performance of the relevant operations in a "cloud computing" environment or as a "software as a service" (SaaS). For example, at least some of the operations may be performed by a group of computers (as examples of machines 500 including processors 504), with these operations being accessible via a network 534 (e.g., the Internet) and via one or more appropriate interfaces (e.g., an API). The performance of certain of the operations may be distributed among the processors, not only residing within a single machine 500, but deployed across a number of machines. In some example implementations, the processors 504 or processor-implemented components may be located in a single geographic location (e.g., within a home environment, an office environment, or a server farm). In other example implementations, the processors 504 or processor-implemented components may be distributed across a number of geographic locations.
[00467] Figure 6 is a block diagram illustrating system 600 that includes an example software architecture 602, which may be used in conjunction with various hardware architectures herein described. Figure 6 is a non-limiting example of a software architecture, and it will be appreciated that many other architectures may be implemented to facilitate the functionality described herein. The software architecture 602 may execute on hardware such as machine 500 of Figure 5 that includes, among other things, processors 504, memory/storage 506, and input/output (I/O) components 508. A representative hardware layer 604 is illustrated and can represent, for example, the machine 500 of Figure 5. The representative hardware layer 604 includes a processing unit 606 having associated executable instructions 608. Executable instructions 608 represent the executable instructions of the software architecture 602, including implementation of the methods, components, and so forth described herein. The hardware layer 604 also includes at least one of memory or storage modules memory/storage 610, which also have executable instructions 608. The hardware layer 604 may also comprise other hardware 612.
[00468] In the example architecture of Figure 6, the software architecture 602 may be conceptualized as a stack of layers where each layer provides particular functionality. For example, the software architecture 602 may include layers such as an operating system 614, libraries 616, frameworks/middleware 618, applications 620, and a presentation layer 622. Operationally, the applications 620 or other components within the layers may invoke API calls 624 through the software stack and receive messages 626 in response to the API calls 624. The layers illustrated are representative in nature and not all software architectures have all layers. For example, some mobile or special purpose operating systems may not provide a frameworks/middleware 618, while others may provide such a layer. Other software architectures may include additional or different layers.
[00469] The operating system 614 may manage hardware resources and provide common services. The operating system 614 may include, for example, a kernel 628, services 630, and drivers 632. The kernel 628 may act as an abstraction layer between the hardware and the other software layers. For example, the kernel 628 may be responsible for memory management, processor management (e.g., scheduling), component management, networking, security settings, and so on. The services 630 may provide other common services for the other software layers. The drivers 632 are responsible for controlling or interfacing with the underlying hardware. For instance, the drivers 632 include display drivers, camera drivers, Bluetooth® drivers, flash memory drivers, serial communication drivers (e.g., Universal Serial Bus (USB) drivers), Wi-Fi® drivers, audio drivers, power management drivers, and so forth depending on the hardware configuration.
[00470] The libraries 616 provide a common infrastructure that is used by at least one of the applications 620, other components, or layers. The libraries 616 provide functionality that allows other software components to perform tasks in an easier fashion than to interface directly with the underlying operating system 614 functionality (e.g., kernel 628, services 630, drivers 632). The libraries 616 may include system libraries 634 (e.g., C standard library) that may provide functions such as memory allocation functions, string manipulation functions, mathematical functions, and the like. In addition, the libraries 616 may include API libraries 636 such as media libraries (e.g., libraries to support presentation and manipulation of various media format such as MPEG4, H.264, MP3, AAC, AMR, JPG, PNG), graphics libraries (e.g., an OpenGL framework that may be used to render two-dimensional and three-dimensional in a graphic content on a display), database libraries (e.g., SQLite that may provide various relational database functions), web libraries (e.g., WebKit that may provide web browsing functionality), and the like. The libraries 616 may also include a wide variety of other libraries 638 to provide many other APIs to the applications 620 and other software components/modules.
[00471] The frameworks/middleware 618 (also sometimes referred to as middleware) provide a higher-level common infrastructure that may be used by the applications 620 or other software components/modules. For example, the frameworks/middleware 618 may provide various graphical user interface functions, high- level resource management, high-level location services, and so forth. The frameworks/middleware 618 may provide a broad spectrum of other APIs that may be utilized by the applications 620 or other software components/modules, some of which may be specific to a particular operating system 614 or platform.
[00472] The applications 620 include built-in applications 640 and third-party applications 642. Examples of representative built-in applications 640 may include, but are not limited to, a contacts application, a browser application, a book reader application, a location application, a media application, a messaging application, or a game application. Third-party applications 642 may include an application developed using the ANDROID™ or IOS™ software development kit (SDK) by an entity other than the vendor of the particular platform and may be mobile software running on a mobile operating system such as IOS™, ANDROID™, WINDOWS® Phone, or other mobile operating systems. The third-party applications 642 may invoke the API calls 624 provided by the mobile operating system (such as operating system 614) to facilitate functionality described herein. [00473] The applications 620 may use built-in operating system functions (e.g., kernel 628, services 630, drivers 632), libraries 616, and frameworks/middleware 618 to create Uls to interact with users of the system. Alternatively, or additionally, in some systems, interactions with a user may occur through a presentation layer, such as presentation layer 622. In these systems, the application/component "logic" can be separated from the aspects of the application/component that interact with a user.
[00474] At least some of the processes described herein can be embodied in computer-readable instructions for execution by one or more processors such that the operations of the processes may be performed in part or in whole by the functional components of one or more computer systems. Accordingly, computer-implemented processes described herein are by way of example with reference thereto, in some situations. However, in other implementations, at least some of the operations of the computer-implemented processes described herein can be deployed on various other hardware configurations. The computer-implemented processes described herein are therefore not intended to be limited to the systems and configurations described with respect to Figures 5 and 6 and can be implemented in whole, or in part, by one or more additional system and/or components.
[00475] Although the flowcharts described herein can show operations as a sequential process, many of the operations can be performed in parallel or concurrently. In addition, the order of the operations can be re-arranged. A process is terminated when its operations are completed. A process can correspond to a method, a procedure, an algorithm, etc. The operations of methods may be performed in whole or in part, can be performed in conjunction with some or all of the operations in other methods, and can be performed by any number of different systems, such as the systems described herein, or any portion thereof, such as a processor included in any of the systems.

Claims

CLAIMS WHAT IS CLAIMED IS:
1 . A method comprising: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a plurality of groups of sequence representations that correspond to a plurality of genomic regions, individual groups of sequence representations of the plurality of groups of sequence representations being aligned with a discrete genomic region; computationally analyzing the individual groups of sequence representations to determine methylation rates of cytosine-guanine dinucleotides included in the individual groups of sequence representations; determining, based on the methylation rates, subsets of sequence representations from among the plurality of groups of sequence representations, individual subsets satisfying one or more methylation rate criteria; generating, based on individual subsets of sequence representations, one or more images for individual genomic regions of the plurality of genomic regions, individual images of the one or more images including a plurality of pixels, wherein individual pixels of the plurality of pixels comprise (i) a first value that corresponds to a genomic location within the individual genomic region, (ii) a second value that corresponds to a methylation rate of sequence representations included in a subset of sequence representations corresponding to the individual genomic region, and (iii) an intensity value indicating a number of the sequence representations included in the subset of sequence representations having the first value and the second value; generating a plurality of convolutional neural networks, individual convolutional networks of the plurality of convolutional networks corresponding to an individual genomic region, wherein the individual convolutional neural networks for the individual genomic regions determine a tumor indication related to a tumor being present in the one or more subjects based on the one or more images that correspond to the individual genomic regions; and computationally analyzing the tumor indications determined by the plurality of convolutional neural networks to determine an overall tumor indication.
2. The method of claim 1 , comprising: obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected; generating first training images based on the first training sequence representations, individual first training images including a first plurality of pixels, wherein individual pixels of the first plurality of pixels comprise (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to methylation rate of at least a portion of the first training sequences representations, and (iii) a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value; obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected; generating second training images based on the second training sequence representations, individual second training images including a second plurality of pixels, wherein individual pixels of the second plurality of pixels comprise (i) a first additional training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second additional training value that corresponds to methylation rate of at least a portion of the second training sequences representation, and (iii) a second intensity training value indicating a number of the second training sequence representations having the first additional training value and the second additional training value; and performing a plurality of iterations of a training process for the individual convolutional neural networks to determine weights of layers of the individual convolutional neural networks.
3. The method of claim 2, wherein individual iterations of the plurality of iterations include: determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the individual convolutional neural networks; determining second weights of layers of the individual convolutional neural networks by providing a second portion of the first training images and a second portion of the second training images to the individual convolutional neural networks; determining differences between the first weights and the second weights; and determining updated weights for the layers of the individual convolutional neural networks based on the differences between the first weights and the second weights.
4. The method of any one of claims 1-3, wherein the methylation rates of cytosine- guanine dinucleotides included in the individual sequence representations are determined by procedures that affect a first nucleobase differently from a second nucleobase.
5. The method of any one of claims 1 -3, comprising: dividing a sample of the one or more samples into a plurality of subsamples including a first subsample corresponding to a first partition and a second subsample corresponding to a second partition, wherein the first partition comprises nucleic acids with a cytosine modification in a greater proportion than additional nucleic acids included in the second partition; wherein the one or more methylation criteria correspond to the nucleic acids included in the first partition or the additional nucleic acids included in the second partition.
6. The method of any one of claims 1 -5, wherein a logistic regression technique is implemented to determine the overall tumor indication based on the tumor indications determined by the plurality of convolutional neural networks.
7. The method of any one of claims 1 -6, wherein individual tumor indications determined by the individual convolutional neural networks include probabilities of a tumor being present in the one or more subjects.
8. The method of claim 7, wherein the probabilities of a tumor being present in the one or more subjects are analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
9. The method of any one of claims 1 -8, wherein: the individual convolutional neural networks include a plurality of output layers with each output layer of the plurality of output layers corresponding to a cancer type of a plurality of cancer types; the individual convolutional neural networks determine a plurality of probabilities of the plurality of cancer types being present in one or more subjects; and for individual output layers, the individual convolutional neural networks generate an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in one or more subjects.
10. The method of claim 9, comprising: computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
11 . The method of any one of claims 1 -10, wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels.
12. The method of any one of claims 1 -10, comprising: determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
13. A system comprising: one or more hardware processors; and memory storing computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a plurality of groups of sequence representations that correspond to a plurality of genomic regions, individual groups of sequence representations of the plurality of groups of sequence representations being aligned with a discrete genomic region; computationally analyzing the individual groups of sequence representations to determine methylation rates of cytosine-guanine dinucleotides included in the individual groups of sequence representations; determining, based on the methylation rates, subsets of sequence representations from among the plurality of groups of sequence representations, individual subsets satisfying one or more methylation rate criteria; generating, based on individual subsets of sequence representations, one or more images for individual genomic regions of the plurality of genomic regions, individual images of the one or more images including a plurality of pixels, wherein individual pixels of the plurality of pixels comprise (i) a first value that corresponds to a genomic location within the individual genomic region, (ii) a second value that corresponds to the methylation rate of sequence representations included in a subset of sequence representations corresponding to the individual genomic region, and (iii) an intensity value indicating a number of the sequence representations included in the subset of sequence representations having the first value and the second value; generating a plurality of convolutional neural networks, individual convolutional networks of the plurality of convolutional networks corresponding to an individual genomic region, wherein the individual convolutional neural networks for the individual genomic regions determine a tumor indication related to a tumor being present in the one or more subjects based on the one or more images that correspond to the individual genomic regions; and computationally analyzing the tumor indications determined by the plurality of convolutional neural networks to determine an overall tumor indication.
14. The system of claim 13, wherein the memory stores additional computer- readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the individual subsets of sequence representations are computational analyzed to generate one or more additional images for the individual genomic regions, wherein: the one or more additional images for the individual genomic regions includes first pixel values that comprise (i) first values that correspond to genomic locations within the individual genomic region and (ii) second values that correspond to an additional molecular characteristic of the individual sequence representations of the group of sequence representations.
15. The system of claim 14, wherein the one or more additional molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the subset of sequence representations corresponding to the individual genomic region, a length of the individual sequence representations of the subset of sequence representations corresponding to the individual genomic region, or a number of restriction enzyme cut sites in the individual sequence representations of the subset of sequence representations corresponding to the individual genomic region.
16. The system of any one of claims 13-15, wherein the one or more methylation rate criteria correspond to at least a threshold number of methylated cytosine-guanine dinucleotides.
17. The system of any one of claims 13-15, wherein the one or more methylation rate criteria correspond to no greater than a threshold number of methylated cytosine- guanine dinucleotides.
18. The system of any one of claims 13-15, wherein the one or more samples are partitioned into a plurality of subsamples on the basis of methylate rate and the one or more methylation rate criteria correspond to a partition of a plurality of partitions into which the plurality of subsamples are divided.
19. A method comprising: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations, the group of sequence representations that correspond to one or more portions of a genomic region; computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations; generating, based on the group of sequence representations, one or more images that include a plurality of pixels, wherein individual pixels of the plurality of pixels comprise (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value; and providing the one or more images to a convolutional neural network, wherein the convolutional neural network computationally analyzes the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects.
20. The method of claim 19, comprising: computationally analyzing the sequencing data to determine a plurality of additional groups of additional sequence representations in relation to a plurality of additional genomic regions; computationally analyzing the plurality of additional groups of additional sequence representations to determine additional values of the one or more molecular characteristics for additional individual sequence representations of the plurality of additional groups of sequence representations; and generating a plurality of additional images based on the plurality of additional groups of sequence representations, wherein: the plurality of additional images include a plurality of additional pixels and individual additional pixels of the plurality of additional pixels comprise (i) an additional first value that corresponds to one or more additional genomic locations, (ii) an additional second value that corresponds to the one or more molecular characteristics, and (iii) an additional intensity value indicating an additional number of the additional sequence representations having the additional first value and the additional second value.
21 . The method of claim 20 wherein: each additional image of the plurality of additional images is generated using information derived from additional sequence representations included in an additional group of sequence representations of the plurality of additional groups of sequence representations and the additional sequence representations are homologous with an additional genomic region.
22. The method of claim 20 or 21 , comprising: providing the plurality of additional images to a plurality of additional convolutional neural networks to determine a plurality of additional tumor indications related to a tumor being present in the one or more samples, wherein individual additional convolutional networks of the plurality of additional convolutional neural networks analyze a portion of the plurality of additional images corresponding to a given genomic region to determine an additional tumor indication related to a tumor being present in the one or more subjects; and computationally analyzing the tumor indication and the plurality of additional tumor indications to determine an overall tumor indication related to a tumor being present in the one or more subjects.
23. The method of claim 22, wherein the tumor indication and the plurality of additional tumor indications are analyzed using a logistic regression technique to determine the overall tumor indication.
24. The method of claim 22 or 23, wherein the tumor indication and the plurality of additional tumor indications include probabilities of a tumor being present in the one or more subjects.
25. The method of claim 24, wherein the probabilities of a tumor being present in the one or more subjects are analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
26. The method of any one of claims 19-25, wherein: the convolutional neural network includes a plurality of output layers with each output layer of the plurality of output layers corresponding to a cancer type of a plurality of cancer types; the convolutional neural network determines a plurality of probabilities of the plurality of cancer types being present in one or more subjects; and for individual output layers, the convolutional neural network generates an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in one or more subjects.
27. The method of claim 26, comprising: computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
28. The method of any one of claims 19-27, wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels.
29. The method of any one of claims 19-28, comprising: determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
30. The method of any one of claims 19-29, wherein: the one or more images include a first image that corresponds to the genomic region and a second image that corresponds to the genomic region; the first image includes first pixel values that comprise (i) first values that correspond to genomic locations within the genomic region and (ii) second values that correspond to a first molecular characteristic of the individual sequence representations of the group of sequence representations; and the second image includes second pixel values that comprise (i) the first values that correspond to the genomic locations within the genomic region and (ii) additional second values that correspond to a second molecular characteristic of the individual sequence representations of the group of sequence representations.
31 . The method of claim 30, comprising: computationally analyzing the first image using a first convolutional neural network to determine a first tumor indication related to a tumor being present in one or more subjects; computationally analyzing the second image using a second convolutional neural network to determine a second tumor indication of a tumor being present in one or more subjects; and determining an overall tumor indication of a tumor being present in one or more subjects based on the first tumor indication and the second tumor indication.
32. The method of any one of claims 19-31 , wherein the one or more molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the group of sequence representations, a length of the individual sequence representations of the group of sequence representations, or a number of restriction enzyme cut sites in the individual sequence representations of the group of sequence representations.
33. The method of any one of claims 19-32, comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
34. The method of any one of claims 19-32, comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having no greater than a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
35. The method of any one of claims 19-34, wherein genomic locations that correspond to first values of the plurality of pixels correspond to an interval that comprises a plurality of nucleotides.
36. The method of any one of claims 19-35, wherein the genomic region is included in a number of genomic regions that are enriched as part of a diagnostic test to determine the presence of tumors in subjects.
37. The method of any one of claims 19-36, comprising: obtaining first training sequence representations derived from first samples obtained from one or more first subjects in which a tumor is detected; generating first training images based on the first training sequence representations, individual first training images including a first plurality of pixels, wherein individual pixels of the first plurality of pixels comprise (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to the one or more molecular characteristics, and (iii) a first intensity training value indicating a number of the first training sequence representations having the first training value and the second training value; obtaining second training sequence representations derived from second samples obtained from one or more second subjects in which a tumor is not detected; generating second training images based on the second training sequence representations, individual second training images including a second plurality of pixels, wherein individual pixels of the second plurality of pixels comprise (i) a first additional training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second additional training value that corresponds to the one or more molecular characteristics, and (iii) a second intensity training value indicating a number of the second training sequence representations having the first additional training value and the second additional training value; and performing a plurality of iterations of a training process for the convolutional neural network to determine weights of layers of the convolutional neural network.
38. The method of claim 37, wherein individual iterations of the plurality of iterations include: determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network; determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network; determining differences between the first weights and the second weights; and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights.
39. A system comprising: one or more hardware processors; and memory storing computer-readable instructions that, when executed by the one or more hardware processors, cause the one or more hardware processors to perform operations comprising: obtaining sequencing data indicating a plurality of sequence representations derived from nucleic acid molecules present in one or more samples obtained from one or more subjects; computationally analyzing the sequencing data to determine a group of sequence representations from among the plurality of sequence representations, the group of sequence representations that correspond to one or more portions of a genomic region; computationally analyzing the group of sequence representations to determine values of one or more molecular characteristics corresponding to individual sequence representations of the group of sequence representations; generating, based on the group of sequence representations, one or more images that include a plurality of pixels, wherein individual pixels of the plurality of pixels comprise (i) a first value that corresponds to a genomic location within the genomic region, (ii) a second value that corresponds to the one or more molecular characteristics, and (iii) an intensity value indicating a number of sequence representations included in the group of sequence representations having the first value and the second value; and providing the one or more images to a convolutional neural network, wherein the convolutional neural network computationally analyzes the one or more images to determine a tumor indication related to a tumor being present in the one or more subjects and the convolutional neural network is trained using first training images generated from first training sequence representations derived from first training subjects in which a tumor is detected and second training images generated from second training sequence representations derived from second training subjects in which a tumor is not detected.
40. The system of claim 39, wherein the memory stores additional computer- readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the sequencing data to determine a plurality of additional groups of additional sequence representations in relation to a plurality of additional genomic regions; computationally analyzing the plurality of additional groups of additional sequence representations to determine additional values of the one or more molecular characteristics for additional individual sequence representations of the plurality of additional groups of sequence representations; and generating a plurality of additional images based on the plurality of additional groups of sequence representations, wherein: the plurality of additional images include a plurality of additional pixels and individual additional pixels of the plurality of additional pixels comprise (i) an additional first value that corresponds to one or more additional genomic locations, (ii) an additional second value that corresponds to the one or more molecular characteristics, and (iii) an additional intensity value indicating an additional number of the additional sequence representations having the additional first value and the additional second value.
41 . The system of claim 40, wherein: each additional image of the plurality of additional images is generated using information derived from additional sequence representations included in an additional group of sequence representations of the plurality of additional groups of sequence representations and the additional sequence representations are homologous with an additional genomic region.
42. The system of claim 40 or 41 , wherein the memory stores additional computer- readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: providing the plurality of additional images to a plurality of additional convolutional neural networks to determine a plurality of additional tumor indications related to a tumor being present in the one or more samples, wherein individual additional convolutional networks of the plurality of additional convolutional neural networks analyze a portion of the plurality of additional images corresponding to a given genomic region to determine an additional tumor indication related to a tumor being present in the one or more subjects; and computationally analyzing the tumor indication and the plurality of additional tumor indications to determine an overall tumor indication related to a tumor being present in the one or more subjects.
43. The system of claim 42, wherein the tumor indication and the plurality of additional tumor indications are analyzed using a logistic regression technique to determine the overall tumor indication.
44. The system of claim 42 or 43, wherein the tumor indication and the plurality of additional tumor indications include probabilities of a tumor being present in the one or more subjects.
45. The system of claim 44, wherein the probabilities of a tumor being present in the one or more subjects are analyzed to determine that a tumor is present in the one or more subjects or that a tumor is absent from the one or more subjects.
46. The system of any one of claims 39-45, wherein: the convolutional neural network includes a plurality of output layers with each output layer of the plurality of output layers corresponding to a cancer type of a plurality of cancer types; the convolutional neural network determines a plurality of probabilities of the plurality of cancer types being present in one or more subjects; and for individual output layers, the convolutional neural network generates an output value including a probability of the plurality of probabilities of an individual cancer type of the plurality of cancer types being present in one or more subjects.
47. The system of claim 46, wherein the memory stores additional computer- readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the plurality of probabilities to determine a type of cancer of the plurality of cancer types having a highest probability of being present in the one or more subjects.
48. The system of any one of claims 39-47, wherein intensity values of the plurality of pixels increases as the number of the sequence representations having the first value and the second value increases; and the intensity values of the plurality of pixels are normalized based on a maximum intensity value for the plurality of pixels.
49. The system of any one of claims 39-48, wherein the memory stores additional computer-readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: determining the intensity value for the individual pixels of the plurality of pixels by determining a logarithmic transformation of a normalized pixel value, the normalized pixel value corresponding to the number of sequence representations having the first value and the second value in relation to the number of sequence representations that are homologous with respect to one or more control genomic regions.
50. The system of any one of claims 39-49, wherein: the one or more images include a first image that corresponds to the genomic region and a second image that corresponds to the genomic region; the first image includes first pixel values that comprise (i) first values that correspond to genomic locations within the genomic region and (ii) second values that correspond to a first molecular characteristic of the individual sequence representations of the group of sequence representations; and the second image includes second pixel values that comprise (i) the first values that correspond to the genomic locations within the genomic region and (ii) additional second values that correspond to a second molecular characteristic of the individual sequence representations of the group of sequence representations.
51. The system of claim 50, wherein the memory stores additional computer- readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing the first image using a first convolutional neural network to determine a first tumor indication related to a tumor being present in one or more subjects; computationally analyzing the second image using a second convolutional neural network to determine a second tumor indication of a tumor being present in one or more subjects; and determining an overall tumor indication of a tumor being present in one or more subjects based on the first tumor indication and the second tumor indication.
52. The system of any one of claims 39-51 , wherein the one or more molecular characteristics include a number of cytosine-guanine dinucleotides present in an individual sequence representation of the group of sequence representations, a length of the individual sequence representations of the group of sequence representations, or a number of restriction enzyme cut sites in the individual sequence representations of the group of sequence representations.
53. The system of any one of claims 39-52, wherein the memory stores additional computer-readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having at least a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
54. The system of any one of claims 39-52, wherein the memory stores additional computer-readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising: computationally analyzing sequencing reads that correspond to the nucleic acid molecules present in the one or more samples to determine a subset of the sequencing reads having no greater than a threshold number of methylated cytosine-guanine dinucleotides present within the genomic region; wherein the sequencing data corresponds to the subset of the sequencing reads.
55. The system of any one of claims 39-54, wherein genomic locations that correspond to first values of the plurality of pixels correspond to an interval that comprises a plurality of nucleotides.
56. The system of any one of claims 39-55, wherein the genomic region is included in a number of genomic regions that are enriched as part of a diagnostic test to determine the presence of tumors in subjects.
57. The system of any one of claims 39-56, wherein: individual first training images include a first plurality of pixels, wherein individual pixels of the first plurality of pixels comprise (i) a first training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second training value that corresponds to the one or more molecular characteristics, and (iii) a first intensity training value indicating a number of first training sequence representations having the first training value and the second training value; individual second training images including a second plurality of pixels, wherein individual pixels of the second plurality of pixels comprise (i) a first additional training value that corresponds to one or more genomic locations within an individual genomic region, (ii) a second additional training value that corresponds to the one or more molecular characteristics, and (iii) a second intensity training value indicating a number of second training sequence representations having the first additional training value and the second additional training value; and the memory stores additional computer-readable instructions, that when executed by the one or more hardware processors, cause the one or more hardware processors to perform additional operations comprising performing a plurality of iterations of a training process for the convolutional neural network to determine weights of layers of the convolutional neural network.
58. The system of claim 57, wherein individual iterations of the plurality of iterations include: determining first weights of layers of the convolutional neural network by providing a first portion of the first training images and a first portion of the second training images to the convolutional neural network; determining second weights of layers of the convolutional neural network by providing a second portion of the first training images and a second portion of the second training images to the convolutional neural network; determining differences between the first weights and the second weights; and determining updated weights for the layers of the convolutional neural network based on the differences between the first weights and the second weights.
PCT/US2025/022042 2024-03-28 2025-03-28 Methods for cancer detection using molecular patterns Pending WO2025208044A1 (en)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
US202463571333P 2024-03-28 2024-03-28
US63/571,333 2024-03-28
US202463651652P 2024-05-24 2024-05-24
US63/651,652 2024-05-24

Publications (1)

Publication Number Publication Date
WO2025208044A1 true WO2025208044A1 (en) 2025-10-02

Family

ID=95399342

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2025/022042 Pending WO2025208044A1 (en) 2024-03-28 2025-03-28 Methods for cancer detection using molecular patterns

Country Status (1)

Country Link
WO (1) WO2025208044A1 (en)

Citations (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20010053519A1 (en) 1990-12-06 2001-12-20 Fodor Stephen P.A. Oligonucleotides
US20030152490A1 (en) 1994-02-10 2003-08-14 Mark Trulson Method and apparatus for imaging a sample on a device
US7537898B2 (en) 2001-11-28 2009-05-26 Applied Biosystems, Llc Compositions and methods of selective nucleic acid isolation
US20110160078A1 (en) 2009-12-15 2011-06-30 Affymetrix, Inc. Digital Counting of Individual Molecules by Stochastic Attachment of Diverse Labels
US8486630B2 (en) 2008-11-07 2013-07-16 Industrial Technology Research Institute Methods for accurate sequence data and modified base position determination
US9150918B2 (en) 2012-06-08 2015-10-06 Pacific Biosciences Of California, Inc. Identifying modified bases using hemi-natural nucleic acids
US9598731B2 (en) 2012-09-04 2017-03-21 Guardant Health, Inc. Systems and methods to detect rare mutations and copy number variation
US20170211143A1 (en) 2014-07-25 2017-07-27 University Of Washington Methods of determining tissues and/or cell types giving rise to cell-free dna, and methods of identifying a disease or disorder using same
US9738894B2 (en) 2003-03-21 2017-08-22 Roche Innovation Center Copenhagen A/S Short interfering RNA (siRNA) analogues
US9850523B1 (en) 2016-09-30 2017-12-26 Guardant Health, Inc. Methods for multi-resolution analysis of cell-free nucleic acids
WO2018009723A1 (en) 2016-07-06 2018-01-11 Guardant Health, Inc. Methods for fragmentome profiling of cell-free nucleic acids
WO2018119452A2 (en) 2016-12-22 2018-06-28 Guardant Health, Inc. Methods and systems for analyzing nucleic acid molecules
US20180216195A1 (en) * 2015-09-17 2018-08-02 The United States Of America, As Represented By The Secretary, Department Of Health And Human Cancer detection methods
US10260088B2 (en) 2015-10-30 2019-04-16 New England Biolabs, Inc. Compositions and methods for analyzing modified nucleotides
US20200185055A1 (en) * 2018-10-12 2020-06-11 Cambridge Cancer Genomics Limited Methods and Systems for Nucleic Acid Variant Detection and Analysis
WO2020160414A1 (en) 2019-01-31 2020-08-06 Guardant Health, Inc. Compositions and methods for isolating cell-free dna
US20200370129A1 (en) 2018-07-23 2020-11-26 Guardant Health, Inc. Methods and systems for adjusting tumor mutational burden by tumor fraction and coverage
US10961525B2 (en) 2017-07-05 2021-03-30 The Trustees Of The University Of Pennsylvania Hyperactive AID/APOBEC and hmC dominant TET enzymes
US20210327534A1 (en) * 2019-12-13 2021-10-21 Grail, Inc. Cancer classification using patch convolutional neural networks
WO2021236778A2 (en) 2020-05-19 2021-11-25 The Trustees Of The University Of Pennsylvania Compositions and methods for dna cytosine carboxymethylation
WO2022197593A1 (en) 2021-03-15 2022-09-22 Illumina, Inc. Detecting methylcytosine and its derivatives using s-adenosyl-l-methionine analogs (xsams)
WO2023288222A1 (en) 2021-07-12 2023-01-19 The Trustees Of The University Of Pennsylvania Modified adapters for enzymatic dna deamination and methods of use thereof for epigenetic sequencing of free and immobilized dna
WO2022226229A9 (en) * 2021-04-21 2023-08-03 Helio Health Inc. Cellular heterogeneity–adjusted clonal methylation (chalm): a methylation quantification method
WO2024073043A1 (en) 2022-09-30 2024-04-04 Illumina, Inc. Methods of using cpg binding proteins in mapping modified cytosine nucleotides

Patent Citations (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20010053519A1 (en) 1990-12-06 2001-12-20 Fodor Stephen P.A. Oligonucleotides
US6582908B2 (en) 1990-12-06 2003-06-24 Affymetrix, Inc. Oligonucleotides
US20030152490A1 (en) 1994-02-10 2003-08-14 Mark Trulson Method and apparatus for imaging a sample on a device
US7537898B2 (en) 2001-11-28 2009-05-26 Applied Biosystems, Llc Compositions and methods of selective nucleic acid isolation
US9738894B2 (en) 2003-03-21 2017-08-22 Roche Innovation Center Copenhagen A/S Short interfering RNA (siRNA) analogues
US8486630B2 (en) 2008-11-07 2013-07-16 Industrial Technology Research Institute Methods for accurate sequence data and modified base position determination
US20110160078A1 (en) 2009-12-15 2011-06-30 Affymetrix, Inc. Digital Counting of Individual Molecules by Stochastic Attachment of Diverse Labels
US9150918B2 (en) 2012-06-08 2015-10-06 Pacific Biosciences Of California, Inc. Identifying modified bases using hemi-natural nucleic acids
US9598731B2 (en) 2012-09-04 2017-03-21 Guardant Health, Inc. Systems and methods to detect rare mutations and copy number variation
US20170211143A1 (en) 2014-07-25 2017-07-27 University Of Washington Methods of determining tissues and/or cell types giving rise to cell-free dna, and methods of identifying a disease or disorder using same
US20180216195A1 (en) * 2015-09-17 2018-08-02 The United States Of America, As Represented By The Secretary, Department Of Health And Human Cancer detection methods
US10260088B2 (en) 2015-10-30 2019-04-16 New England Biolabs, Inc. Compositions and methods for analyzing modified nucleotides
WO2018009723A1 (en) 2016-07-06 2018-01-11 Guardant Health, Inc. Methods for fragmentome profiling of cell-free nucleic acids
US9850523B1 (en) 2016-09-30 2017-12-26 Guardant Health, Inc. Methods for multi-resolution analysis of cell-free nucleic acids
WO2018119452A2 (en) 2016-12-22 2018-06-28 Guardant Health, Inc. Methods and systems for analyzing nucleic acid molecules
US10961525B2 (en) 2017-07-05 2021-03-30 The Trustees Of The University Of Pennsylvania Hyperactive AID/APOBEC and hmC dominant TET enzymes
US20200370129A1 (en) 2018-07-23 2020-11-26 Guardant Health, Inc. Methods and systems for adjusting tumor mutational burden by tumor fraction and coverage
US20200185055A1 (en) * 2018-10-12 2020-06-11 Cambridge Cancer Genomics Limited Methods and Systems for Nucleic Acid Variant Detection and Analysis
WO2020160414A1 (en) 2019-01-31 2020-08-06 Guardant Health, Inc. Compositions and methods for isolating cell-free dna
US20210327534A1 (en) * 2019-12-13 2021-10-21 Grail, Inc. Cancer classification using patch convolutional neural networks
WO2021236778A2 (en) 2020-05-19 2021-11-25 The Trustees Of The University Of Pennsylvania Compositions and methods for dna cytosine carboxymethylation
WO2022197593A1 (en) 2021-03-15 2022-09-22 Illumina, Inc. Detecting methylcytosine and its derivatives using s-adenosyl-l-methionine analogs (xsams)
WO2022226229A9 (en) * 2021-04-21 2023-08-03 Helio Health Inc. Cellular heterogeneity–adjusted clonal methylation (chalm): a methylation quantification method
WO2023288222A1 (en) 2021-07-12 2023-01-19 The Trustees Of The University Of Pennsylvania Modified adapters for enzymatic dna deamination and methods of use thereof for epigenetic sequencing of free and immobilized dna
WO2024073043A1 (en) 2022-09-30 2024-04-04 Illumina, Inc. Methods of using cpg binding proteins in mapping modified cytosine nucleotides

Non-Patent Citations (67)

* Cited by examiner, † Cited by third party
Title
"Genetics Computer Group", UNIVERSITY RESEARCH PARK, article "Wisconsin Sequence Analysis Package"
ALTSCHUL ET AL., J. MOL. BIOL., vol. 215, 1990, pages 403 - 410
BELINKSY, ANNU. REV. PHYSIOL, vol. 77, 2015, pages 453 - 74
BOCK ET AL., NAT BIOTECH, vol. 28, 2010, pages 1106 - 1114
BOOTH ET AL., SCIENCE, vol. 336, 2012, pages 934 - 937
BROWN: "Genomes", 2002, JOHN WILEY & SONS, INC
CUDDAPAH ET AL., GENOME RES, vol. 19, 2009, pages 24 - 32
D.A. MOSER ET AL.: "Targeted bisulfite sequencing: A novel tool for the assessment of DNA methylation with high sensitivity and increased coverage", PSYCHONEUROENDOCRINOLOGY, vol. 120, 2020, pages 1 - 8, XP086264386, DOI: 10.1016/j.psyneuen.2020.104784
EHRLICH, EPIGENOMICS, vol. 1, 2009, pages 239 - 259
FOX-FISHER ET AL., ELIFENOV, vol. 29, 2021, pages 10
FREIER, NUCLEIC ACIDS RES., vol. 25, 1997, pages 4429 - 4443
FURONAKA ET AL., PATHOLOGY INTERNATIONAL, vol. 55, 2005, pages 303 - 309
GALE ET AL., PLOS ONE, vol. 13, 2018, pages 0194630
GANSAUGE ET AL., NATURE PROTOCOLS, vol. 8, 2013, pages 737 - 748
GOMES ET AL., REV. PORT. PNEUMOL, vol. 20, 2014, pages 20 - 30
GREER ET AL., CELL, vol. 161, 2015, pages 868 - 878
GUO ET AL., CLIN. CANCER RES., vol. 10, 2004, pages 7917 - 24
GUO ET AL., NAT. COMMUN, vol. 9, 2018, pages 1520
HAN ET AL., MOL. CELL, vol. 63, 2016, pages 711 - 719
HELLER ET AL., ONCOGENE, vol. 25, 2006, pages 959 - 968
HON ET AL., GENOME RES, vol. 22, 2012, pages 246 - 258
HOPKINS-DONALDSON ET AL., CELL DEATH DIFFER., vol. 10, 2003, pages 356 - 64
HULBERT ET AL., CLIN. CANCER RES., vol. 23, 2017, pages 1998 - 2005
KANG ET AL.: "Cancer Genome Atlas", GENOME BIOLOGY, vol. 18, 2017, pages 53
KANG, GENOME BIOL, vol. 18, 2017, pages 53
KATAINEN ET AL., NATURE GENETICS, 8 June 2015 (2015-06-08)
KIKUCHI ET AL., CLIN. CANCER RES, vol. 11, 2005, pages 2954 - 61
KIM ET AL., ONCOGENE, vol. 20, 2001, pages 1765 - 70
KINDE ET AL., PROC NAT'LACAD SCI USA, vol. 108, 2011, pages 9530 - 9535
KOU ET AL., PLOS ONE, vol. 11, 2016, pages 0146638
KUMAR ET AL., FRONTIERS GENET, vol. 9, 2018, pages 640
LAM ET AL., BIOCHIM BIOPHYS ACTA, vol. 1866, 2016, pages 106 - 20
LEIT5O ET AL.: "Locus-specific DNA methylation analysis by targeted deep bisulfite sequencing", METHODS MOL BIOL, vol. 1767, no. 351-66, pages 20 - 18
LI, HDURBIN, R: "Fast and accurate short read alignment with Burrows-Wheeler transform", BIOINFORMATICS, vol. 25, no. 14, 2009, pages 1754 - 1760
LICCHESI ET AL., CARCINOGENESIS, vol. 29, 2008, pages 895 - 904
LISSA ET AL., TRANSL LUNG CANCER RES, vol. 5, no. 5, 2016, pages 492 - 504
LIU ET AL., NAT CHEM BIOL, vol. 13, 2017, pages 181 - 187
LIU ET AL., NAT CHEM BIOL., vol. 13, no. 2, pages 181 - 187
LIU ET AL., NATURE BIOTECHNOLOGY, vol. 37, 2019, pages 424 - 429
LOYFER ET AL., BIORXIV, 2022, Retrieved from the Internet <URL:https://doi.org/10.1101/2022.01.24.477547>
MARTIN ET AL., NAT. STRUCT. MOL. BIOL, vol. 18, 2011, pages 708 - 14
MEISSNER, AGNIRKE, ABELL, G.WRAMSAHOYE, BLANDER, E.SJAENISCH, R: "Reduced representation bisulfite sequencing for comparative high-resolution DNA methylation analysis", NUCLEIC ACIDS RESEARCH, vol. 33, 2005, pages 5868 - 5877, XP002661907, DOI: 10.1093/nar/gki901
MOSS ET AL., NAT COMMUN, vol. 9, 2018, pages 5068
NEEDLEMANWUNSCH, J. MOL. BIOL., vol. 48, 1970, pages 443 - 453
OOKI ET AL., CLIN. CANCER RES, vol. 23, 2017, pages 7141 - 52
PALMISANO ET AL., CANCER RES., vol. 63, 2003, pages 4620 - 4625
PARDOLL, NATURE REVIEWS CANCER, vol. 12, 2012, pages 252 - 264
RHEE ET AL., CELL., vol. 147, 2011, pages 1408 - 19
SAMBROOK ET AL.: "Molecular Cloning, A Laboratory Manual", 1989, COLD SPRING HARBOR LABORATORY PRESS
SCHATZ, NATURE METHODS., vol. 14, no. 4, 2017, pages 347 - 348
SCHNEIDER ET AL., BMC CANCER, vol. 11, 2011, pages 102
SCHUTSKY ET AL., NATURE BIOTECHNOLOGY, vol. 36, 2018, pages 1083 - 1090
SEVERIN, NUCLEIC ACIDS RES., vol. 39, 2011, pages 8740 - 8751
SHAREEF, S.JBEVILL, S.MRAMAN, A.T ET AL.: "Extended-representation bisulfite sequencing of gene regulatory elements in multiplexed samples and single cells", NAT BIOTECHNOL, vol. 39, 2021, pages 1086 - 1094, XP037559958, DOI: 10.1038/s41587-021-00910-x
SHI ET AL., BMC GENOMICS, vol. 18, 2017, pages 901
SKVORTSOVA ET AL., BR. J. CANCER, vol. 94, no. 10, 2006, pages 1492 - 1495
SNYDER ET AL., CELL, vol. 164, 2016, pages 57 - 68
SONG ET AL., NAT BIOTECH, vol. 29, 2011, pages 68 - 72
SUN ET AL., BIOESSAYS, vol. 37, 2015, pages 1155 - 62
T. GONG ET AL.: "Analysis and performance assessment of the whole genome bisulfite sequencing data workflow: currently available tools and a practical guide to advance DNA methylation studies", SMALL METHODS, vol. 6, no. 21, 2022, pages 01251
TOYOOKA ET AL., CANCER RES., vol. 61, 2001, pages 4556 - 4560
VAISVILA ET AL.: "Discovery of novel DNA cytosine deaminase activities enables a nondestructive single-enzyme methylation sequencing method for base resolution high-coverage methylome mapping of cell-free and ultra-low input DNA", BIORXIV, 2023, Retrieved from the Internet <URL:https://www.biorxiv.org/content/10.1101/2023.06.29.547047v1>
VAISVILA R ET AL.: "EM-seq: Detection of DNA methylation at single base resolution from picograms of DNA", BIORXIV, 2019, Retrieved from the Internet <URL:www.biorxiv.org/content/10.1101/2019.12.20.884692v1>
YAMASHITA ET AL., NUCLEIC ACIDS RES., 2006, pages 86 - 89
YANG ET AL., BIO-PROTOCOL, vol. 12, no. 17, 2023, pages 4496
YU ET AL., CELL, vol. 149, 2012, pages 1368 - 80
ZHANGMADDEN, GENOME RES, vol. 7, 1997, pages 649 - 656

Similar Documents

Publication Publication Date Title
JP7573536B2 (en) Compositions and methods for isolating cell-free DNA
JP7717057B2 (en) Compositions and methods for analyzing cell-free DNA in methylation partitioning assays
US20220344004A1 (en) Detecting the presence of a tumor based on off-target polynucleotide sequencing data
WO2023197004A1 (en) Detecting the presence of a tumor based on methylation status of cell-free nucleic acid molecules
JP2025542260A (en) Methods involving methylation-preserving amplification with error correction
US20250101494A1 (en) Methods for analyzing cytosine methylation and hydroxymethylation
US20250078955A1 (en) Detecting the presence of a tumor based on methylation status of cell-free nucleic acid molecules
JP2025522763A (en) Enrichment of aberrantly methylated DNA
JP2024523401A (en) Methods and compositions for copy number information-based tissue origin analysis
JP2023524681A (en) Methods for sequencing using distributed nucleic acids
JP2025542261A (en) Integrated targeted and whole-genome somatic and DNA methylation sequencing workflow
WO2024233502A1 (en) Cell-free dna blood-based test for cancer screening
WO2024073508A2 (en) Methods and compositions for quantifying immune cell dna
WO2025208044A1 (en) Methods for cancer detection using molecular patterns
EP4314329A1 (en) Methods and compositions for quantifying immune cell dna
WO2025076452A1 (en) Detecting tumor-related information based on methylation status of cell-free nucleic acid molecules
WO2025250656A1 (en) Machine learning classification model for cancer detection
WO2025064706A1 (en) Detecting the presence of a tumor based on methylation status of cell-free nucleic acid molecules
US20250336491A1 (en) Machine learning models to test computational algorithms
US20260009024A1 (en) Non-invasive monitoring of genomic alterations induced by gene-editing therapies
US20240412815A1 (en) Detecting homologous recombination deficiences based on methylation status of cell-free nucleic acid molecules
US20250101522A1 (en) Brca1 promoter methylation in sporadic breast cancer patients detected by liquid biopsy
US20250243550A1 (en) Minimum residual disease (mrd) detection in early stage cancer using urine
CN118974279A (en) Detecting the presence of tumors based on the methylation status of cell-free nucleic acid molecules
WO2025235602A1 (en) Predictive, prognostic signatures for immuno-oncology using liquid biopsy

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 25719248

Country of ref document: EP

Kind code of ref document: A1