CN106202989A - A kind of method obtaining child's individuality biological age based on oral microbial community - Google Patents

Abstract

The present invention provides a kind of method obtaining child's individuality biological age based on oral microbial community, and described method includes obtaining the sample containing described child individuality oral microorganism；Extract the DNA of oral microorganism；Described DNA information is converted into microbiologic population's information, utilizes random forests algorithm, oral microbial community information and age are carried out regression analysis, build regression model, it is thus achieved that described Chinese population child's Individual Age.The scheme that the present invention provides can obtain the biological age that Chinese population child is individual exactly, can be without invasive, obtain saliva of buccal cavity or dental plaque sample simply, efficiently, child's Individual Age is detected for a long time, this is beneficial to quickly judge host's now physiological health state, give a clue for health monitoring, improve disease early diagnosis speed simultaneously.

Description

A method to obtain individual biological age of children based on oral microflora

technical field

The invention relates to the field of microbial detection models, in particular to a method for obtaining the biological age of children based on oral microbial communities.

Background technique

Human beings are not alone in the world. Each human body carries billions of microorganisms. Human beings and the symbiotic microorganisms in their bodies together form a "super organism". There are no microorganisms in the womb, and the first contact with microorganisms is the birth canal. After birth, more microbes migrate into the human body through drinking milk and contact with the external environment. The human microbial community has age characteristics. The microbial community in the human body is gradually established with age and continues to evolve with changes in physiological development throughout life. Those microorganisms that enter the human body after birth and have an important impact on human health are important carriers of acquired endowments, which is equivalent to the existence of another genome in the human body besides the human genome that regulates the life and health of the human body through expression. It is currently believed that Symbiotic microorganisms can be used as the second genome of the human body, and the sum of their genetic information is called the microbiome, endowing humans with complex individual characteristics that do not depend on their own evolution. Therefore, a comprehensive understanding of the human symbiotic flora can deeply reveal its impact on human health or disease status, thereby establishing the relationship between the existence and changes of the microbial community and the physiological state of the host.

The oral system is a transportation hub that connects the inside and outside of the human body, and is a very important habitat for the symbiotic flora of the human body. Maintaining a healthy balance of the structure and function of the oral flora is of profound and significant significance to human health. Compared with blood tests and bone age as a medium for disease diagnosis, oral site sampling has the advantages of low invasiveness, low cost, simple and fast sample collection and processing.

Human growth and development can be represented by two "ages": biological age and chronological age. Biological age refers to the current position of an individual in the potential life cycle, is a comprehensive index of human health status, and is an objective expression of the aging degree of the body. Life age refers to the actual age of an individual calculated from the date of birth, based on the elapsed time on the calendar. Due to the influence of nutrition, disease, genetics, environment and other factors, some people's life age and developmental level (biological age) are not consistent, so life age can not truly reflect the development and maturity of a body, and biological age is related to an individual's physiological Health is closely related.

Currently, skeletal age (bone age; skeletal age, SA; bone age, BA) is widely used to evaluate biological age or maturity status. Bone age measurement includes body parts such as the wrist, elbow joint, knee joint, and foot. Because of the sensitivity of the wrist and the convenience of taking pictures, X-ray films are often used in clinical practice to measure the wrist of children. But there are many bones in the wrist, including 8 carpal bones, 5 metacarpal bones, 14 phalanx bones, and 29 ulna and radius bones. In addition, the seed bone on the inner side of the thumb is also an important sign of bone development. Evaluation methods mainly include atlas method and scoring method. Among them, the atlas method is simple and intuitive, but the appraiser still compares and interprets the whole X-ray film in actual operation, and there are still problems such as strong subjective interpretation and many bone maturation combinations; Although the scoring method is relatively objective, the division of bone development grades is too detailed and the standards are difficult to grasp, which reduces the reliability of bone age evaluation. Moreover, the establishment of a bone age standard atlas library and the research on computer image reading systems need to be solved urgently. In addition, although X-rays are almost harmless to the human body, long-term follow-up of children's growth and development still requires protective measures during treatment, and the development of non-radiographic bone age development evaluation and methods is still in the early stage, such as the interpretation accuracy of ultrasonic testing. However, there are still methodological problems. In addition to bone age, the commonly used methods for determining biological age are tooth maturity and secondary sexual characteristic development. However, these methods usually rely on the subjective judgment of the appraiser, and the results are all range values. age, and the evaluation indicators are relatively heterogeneous among individuals. Therefore, it is urgent to develop an objective, accurate, easy-to-operate, noninvasive, and high-throughput biological age assessment method.

Contents of the invention

In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is to provide a method for obtaining the individual biological age of children based on oral microbial flora.

The technical solution adopted by the present invention to achieve the above object is: a method for obtaining the biological age of children based on the oral microbial community, comprising the following steps:

Data Collection: Collect individual oral samples from children at multiple time points;

Data conversion: extract the DNA information of oral samples, and use bioinformatics methods to convert the DNA information into oral microbial community information;

Preliminary construction of the data model: the obtained oral microbial community information is used as an input variable, and the random forest method is used to regress it on the age information, and a preliminary mathematical model for detecting biological age based on the oral microbial community information is initially constructed;

Optimization and determination of the mathematical model: according to the order of the importance of variables in the model, the combination of model variables is simplified without affecting the performance of the model, and finally the model for individual age detection of children is determined;

Detection of individual biological age of children: The required microbial community information is used as an input variable, and the established mathematical model is used for regression analysis to obtain the biological age of individual children detected at this time.

The oral cavity sample is a sample of saliva or supragingival plaque.

Said converting DNA information into oral microbial community information comprises the following steps:

Obtain 16s RNA or whole genome information of DNA information through high-throughput sequencing;

According to the 16s RNA or the whole genome information, the bacterial germline information is assigned from the phylum to the species level;

The sequence number of each species at the species level was counted separately for each sample, and its ratio was calculated with the sequence number obtained in the sample as a whole, so as to obtain the relative abundance of each species.

The preliminary construction of the data model includes the following steps:

1) The composition and relative abundance of all bacterial species levels of the obtained oral microorganisms are used as input variables;

2) Using the random forest method, the input variables are regressed on the age information of individual children, and a preliminary mathematical model for detecting biological age based on oral microbial community information is initially constructed.

The optimization and determination of the data model includes the following steps:

1) Obtain the degree of importance of each variable representing the type of bacteria in the preliminary mathematical model to the performance of the model;

2) According to the importance of variables to the model, sort them from small to large, gradually reduce the number of variables, and use the random forest method to perform regression analysis on age to obtain models with different variable combinations;

3) Evaluate the most simplified variable combination without reducing the performance of the model, and determine it as an age-related variable, so as to determine the final optimization model.

The detection of the individual biological age of the child comprises the following steps:

1) Obtain DNA from individual oral samples of children;

2) Using bioinformatics methods to convert DNA information into oral microbial community information;

3) Obtain the relative abundance of age-related variables of individual children;

4) Using the random forest method, the composition and abundance of age-related variables are used as variables, and regression analysis is performed on the established age detection model to obtain the biological age of individual children at this time.

It also includes: comparing the biological age of the obtained individual child with its actual age, and knowing the growth and development of the child at this time, that is, if the biological age is lower than the actual age, it indicates that the child may be stunted due to diseases and other factors ; If the biological age is equal to the actual age, it indicates that the child's growth and development are normal; if the biological age is higher than the actual age, it indicates that the child may have precocious puberty.

The present invention has the following advantages and beneficial effects:

1. The object collection and processing of the present invention are simple, non-invasive, and low in cost;

2. The model establishment and optimization of the present invention are easy to operate and efficient in data processing;

3. The evaluation of the present invention is objective and automatic, and can provide accurate numerical values;

4. The present invention is widely used: its application object is not only suitable for large-scale crowd assessment, but also can realize long-term monitoring for individuals; its application form can not only detect the biological age of children at this time, but also can be used as a tool for evaluating individual development, growth and health. helper method.

Description of drawings

Fig. 1 is the experimental design diagram that the present invention implements to provide;

Fig. 2 is a characteristic figure of oral microbial community structure provided by the implementation of the present invention;

Fig. 3 screens out age-related oral microbial composition and its contribution to model performance by the random forest regression method provided by the present invention;

Fig. 4 is a graph showing the results of applying the optimized model provided by the present invention to the healthy group and caries group.

detailed description

The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments.

The present invention uses oral dental plaque and saliva microbial flora to construct and optimize the biological age of the detectable children's oral cavity as an example (Fig. 1), including the following:

(1) Collect the clinical information of children's oral health status (Table 1):

A follow-up survey was conducted on the oral health of full-time children in Guangzhou Nanfang Chinese-English Kindergarten. The oral health was checked every six months, three times a year, and then checked at intervals of one year. According to the purpose of this study, children with the following three types of oral health changes were selected to be included in this research: ①Healthy group (H2H group): 17 children whose oral caries status remained healthy all the time; ②Caries group, including dental caries Disease occurrence group (H2C group): 21 children whose oral caries status has gone through the process from healthy to new caries, and caries progress group (C2C group): oral caries status has gone through the development process from caries to caries of 12 children. Inclusion criteria include: age about 4 years old, all 20 deciduous teeth have erupted. Exclusion criteria include: systemic diseases, oral diseases such as periodontal and bad breath, and taking antibiotics for three months. The consent of the guardians of the volunteers was obtained for the details of the entire experimental process and subsequent data release, and an informed consent was signed. A total of 284 samples of supragingival plaque and saliva were taken during the oral examination of all selected children.

Investigation method: Two dental endodontic specialists inspected by inspection combined with probing. The inspection instruments were sterilized by high temperature and high pressure, and soft dirt was removed with cotton swabs if necessary. Before the inspection, the understanding, methods and standards were unified, and the Kappa values of the standard consistency test were all greater than 0.92. The diagnostic criteria for caries were adopted from the World Health Organization "Basic Methods of Oral Health Survey" (1997). Diagnostic criteria for crown caries: tooth pits and fissures or smooth surfaces have obvious caries, or obvious subenamel damage, or lesions that can be clearly explored and soften the bottom or wall of the cavity are recorded as dental caries, including fillings or pits Groove closed at the same time caries. The following manifestations without other positive symptoms are not included in the scope of dental caries records: ① white or chalky spots; ② no softening staining or rough spots on probing; ③ enamel gaps or pits and fissures staining, but no obvious subenamel damage ; ④ Moderate to severe dental fluorosis, shiny, hard, with small depressions; ⑤ According to the distribution or medical history, combined with palpation and inspection, observe the caries caused by wear and tear.

The sample clinical data that table 1 example of the present invention provides

(2) Collect children's saliva and supragingival plaque samples:

The subjects refrained from eating and drinking one hour before the sampling, and each sampling was between 9:100 and 12:00 in the morning. During the sampling, the children kept their heads lightly, eyes closed, and sat upright. Collect about 3-5ml of non-irritating saliva from children in a 50ml sterile centrifuge tube, and divide each 1ml into a 1.5ml centrifuge tube; then use a sterile toothbrush to collect all the plaques on the erupted primary gums for 1 minute, and the adherent The plaque on the toothbrush was transferred to a 50ml centrifuge tube filled with 10ml of double distilled water. Avoid touching the mucous membrane and other oral sites when sampling. The samples were numbered and placed at -80°C to store the DNA to be extracted.

(3) Genomic DNA extraction and PCR amplification of 16S rRNA gene fragments

A high-salt DNA extraction method was used. Centrifuge the centrifuge tubes containing plaque and saliva at 13,000 rpm/min for 15 min, discard the supernatant, add 1 ml of lysate, add 30 μL of proteinase K and 150 μL of 10% SDS to the lysate mixture, and culture overnight at 53°C with shaking in a water bath. Add 400 μL of 5M NaCl and incubate on ice for 10 min, then centrifuge at 13,000 rpm/min for 10 min. Add an equal volume of saturated phenol solution until the phenol in the aqueous phase is mixed into an emulsion, centrifuge at 13,000 rpm/min for 15 min, absorb the upper viscous aqueous phase into a new tube, and repeat the phenol extraction once. Add an equal volume of chloroform-isoamyl alcohol mixture (24:1), rotate and mix well, centrifuge at 13,000rpm/min for 15min, and transfer the upper viscous aqueous phase. Add 800 μL of isopropanol, incubate at room temperature for 1 min, and centrifuge at 13,000 rpm/min for 15 min. Discard the supernatant, wash twice with 70% ethanol, dry and dissolve in 50 μL TE solution.

The DNA concentration was quantified by a Qubit ultra-micro spectrophotometer, and the DNA integrity was detected by electrophoresis. The extracted DNA was stored at -20°C. About 15ng of DNA was used to construct the 16S amplified library.

In order to obtain relatively accurate phylogenetic information, the V1-V3 hypervariable region (Escherichia coli positions 5-534) on the 16S rRNA fragment was selected as the target fragment for PCR amplification. Determine the PCR upstream primer (5'-NNNNNNN-TGGAGAGTTTGATCCTGGCTCAG-3') and downstream primer (5'-NNNNNNNN-TACCGCGGCTGCTGGCAC-3'), NNNNNNN is the IDtag, which is a random combination of seven bases designed to distinguish different sample sources , add the 5' ends of the upstream and downstream primers respectively, and use the multi-sample parallel labeling technology to complete the sequencing of multiple samples on the sequencer at the same time.

Each sample was subjected to PCR amplification three times. The PCR reaction system (25 μL) contained 12.5 μL of Gotag Hotstart polymerase, 1 μL of upstream and downstream primers (concentration 5 pM), 1 μL of genomic DNA (5 ng μL-1), 9.5 μL of PCR-grade sterile water , and reacted in the Thermocycler PCR system. The reaction conditions were set as follows: pre-denaturation at 95°C for 2 min, denaturation at 94°C for 30 s, annealing at 56°C for 25 s, extension at 72°C for 25 s, a total of 25 cycles, and finally extension at 72°C for 5 min. After the PCR products were mixed, all were subjected to gel electrophoresis (1.2% Q agarose, 5V cm-1, 40min) to confirm the amplification effect, and the agarose gel was placed under a UV lamp to cut out a DNA band of about 500bp in length. The operating procedures provided by the Qiagen MiniElute kit were used to recover and purify the DNA of the target fragment, and wash with 20 μL.

(4) 454GS FLX Titanium Sequencing

The main process is as follows: ①Library preparation, using Agilent BioAnalyzer 2100 bioanalyzer and PicoGreen ultra-micro spectrophotometer for joint quantification, different samples were mixed in equimolar amounts to construct three DNA libraries, connected with specific adapters for modification, denatured and recovered Single-stranded DNA; ②Emulsified PCR, immobilizing the DNA library on magnetic beads, emulsified by amplification, to form an oil-water mixture, and each DNA fragment is independently amplified in parallel in a microreactor to produce millions of identical copies. Break the emulsified state, recover and purify the DNA fragments bound to the magnetic beads; ③sequencing reaction, mix the magnetic beads carrying the DNA with other reactants, put them in a PTP plate and place them in a 454GS FLX Titanium machine, each of which is complementary to the template strand The addition of nucleotides will generate fluorescent signals and be captured by the CCD camera, and the sequencing will be completed step by step; ④ data collection, the base analysis of the sequencing reaction data will be performed through system informatics tools.

(5) Convert the obtained high-throughput data into specific microbial community data

Sequence quality control: The 454 high-quality sequence analysis process is mainly based on the MOTHUR platform, and quality control specifications are set. Sequence fragments that meet the standards are regarded as high-quality sequences and are retained. ①At least one end of the primer can be matched, and the allowed editing distance (number of bases for insertion, deletion, deletion, mismatch) does not exceed 2; ②The sequence length is greater than 150bp; ③Set a 50bp base reading frame, from each The first base of the sequence starts to move backward base by base, and calculates the average quality score in the reading frame every time a base is moved, and the average quality index must be greater than 35; ④ does not contain ambiguous bases; ⑤ allow tags The number of sequence mismatches does not exceed 1. After preliminary filtering, the sequence needs to be further screened for sequencing errors, including steps such as "preclustering" and chimera sequence search. Select the UCHIME program to find and delete these sequences.

Analysis of phylogenetic information based on 16S database: use the MOTHUR classification method to classify bacterial phylogenetic information from phylum to species level in the 16S database (CORE) of human oral core microorganisms, and count the phylogenetic information of each sample at each classification level Sequence number, and calculate its ratio with the total sequence number obtained in the sample, so as to obtain the relative abundance of each species of each phylum.

(6) The influence of different factors on the distribution of oral flora (Figure 2):

Community structure calculation method based on Jensen-Shannon matrix: In addition to the evolutionary distance between samples, it can also investigate the difference in the abundance of bacterial species in samples. The abundance distribution of bacterial species in a sample can be regarded as the probability distribution of species, and the mutual information entropy (Jensen-Shannon divergence, JSD) of this probability distribution between samples can be used to measure the difference of microbiome between samples. The formula for calculating the distance D(a,b) between samples is as follows:

$\mathrm{<span\; class="notranslate">\; D.</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; b</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\sqrt{\mathrm{<span\; class="notranslate">\; JSD</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; a</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; )</span>}$

P _a and P _b represent the abundance distribution in sample a and sample b, respectively. JSD(X,Y) defines the mutual information entropy (Jensen-Shannon divergence) between different probability distributions X and Y in two samples.

$\mathrm{<span\; class="notranslate">\; JSD</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; KLD</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; KLD</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; )</span>$

$\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span>$

KLD is the Kullback-Leibler dispersion between X and Y, the specific calculation method is as follows:

$\mathrm{<span\; class="notranslate">\; KLD</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\mathrm{<span\; class="notranslate">\; log</span>}\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{{\mathrm{<span\; class="notranslate">\; Y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}$

Unsupervised Principal Coordinates Analysis: Principal Coordinates Analysis (PCoA: Principal Coordinates Analysis) is performed on the Jensen-Shannon matrix to show the structural characteristics of oral microbial communities between different samples. × Variable relative abundance matrix is analyzed to observe the internal flora results of the sample unbiased and overall without considering the influence of environmental factors, and find one or more potential variables (principal coordinates, Principal coordinate, PC ) to best explain the internal variation of the sample in the lower dimension to the greatest extent, and each principal coordinate represents the degree of overall structural variation that can be explained in this dimension, so as to achieve the purpose of data dimensionality reduction and sample sorting, among which The sample score (Score) is a linear combination of species scores.

The results of permutation multivariate statistical analysis showed that the oral microbial community had obvious age characteristics, which were related to individual developmental maturity and health status, and supported the establishment of a method for assessing the biological age of children based on the oral microbial community (Fig. 1, Fig. 2):

①In each ecological site, the time/age factor was the most important factor determining the distribution of the flora.

②In each ecological site, other important factors affecting the distribution of flora were ranked according to their importance: health/disease status, sample grouping, and individual heterogeneity.

③In different groups (including H2H, H2C, C2C groups), the time factor in the healthy group had the greatest impact on its flora, while in the caries group, the time factor was affected by the disease state and its effect on the flora was inhibited.

The above results suggest that oral flora can be used as a medium for individual age detection and can reflect the oral health status of the host.

(7) Preliminary establishment of a mathematical model for oral state detection (Fig. 1)

In machine learning, the random forest method is a model that contains multiple decision trees, and its output category is determined by the mode of the category output by individual trees. This model is widely used to mine the relationship between the target variable and many explanatory variables. relationship. This method can not only establish a classification or regression model, but also determine the variables that distinguish a specific state or label, and judge the size of its distinguishing ability through its importance value. In this example, the random forest method is implemented using the randomForest package of R, and 5000 trees are established, and the others are default settings. 2/3 of the input data is used as the training data set, and 1/3 of the input data is used as the test data set, and 100 experiments are randomly performed to reduce the error.

Taking the bacterial species data of the oral microbial community in the H2H group as the input variable, taking the actual month age corresponding to each sample as the sample information, and regressing it to the discrete output variable (predicted month age), a preliminary establishment of a method for detecting the individual biological age of children mathematical model.

Random forest machine learning (Random Forests, RF) is a kind of machine learning based on the classifier algorithm, proposed by Leo Breiman, through the self-help method resampling technology, from the training set (data set) n Repeated random sampling k The sample generates a new set of training samples (train set), and then generates k classification trees to form a random forest based on the self-service sample set. The classification result of new data depends on the score formed by the number of votes of the classification tree, and the classification error depends on each tree. classification ability and the correlation between them. The classification ability of a single tree may be very small, but after a large number of decision trees are randomly generated, a test sample can select the most probable classification through statistics of the classification results of each tree. It improves the prediction accuracy of the model by summarizing a large number of classification trees. Because it does not have overfitting and high prediction accuracy, this model is widely used to mine the relationship between the target variable and many explanatory variables.

(8) Optimizing the established model for detecting the individual biological age of children (Fig. 3)

In addition to building detection models and predictions, the random forest method can also be used to evaluate the importance of explanatory variables. Feature selection uses a random method to split each node, and then compares the errors generated in different situations. The intuitive evaluation standard is that the more important the variable is, the greater the impact on the forecast results. The evaluation of the importance of the random forest model explanatory variables adopts a similar standard: the value of a certain explanatory variable of all test samples is randomly disrupted, and the original random forest model is used to predict the test samples again. The more the out-of-bag fitting error increases, the explanation Variables are more important. The increase of out-of-bag fitting error can be used to quantitatively evaluate the importance of explanatory variables. This patent uses Ten-Fold Cross Validation (Ten-Fold Cross Validation) to evaluate the minimum number of variables required to build a model. Repeat 100 times randomly, and use the mean value as an estimate of the accuracy of the algorithm. Cross-Validation (CV) is a statistical analysis method used to verify the performance of classifiers, mainly used to obtain reliable and stable models in modeling evaluation, that is, to group the original data (dataset) in a certain sense , one part is used as the training set (train set), and the other part is used as the validation set (validation set). First, the training set is used to train the classifier, and then the validation set is used to test the trained model, and each classification error is taken as A metric for evaluating classifier performance. The ten-fold cross-validation divides the data set into ten parts, and takes turns using 9 parts as training data and 1 part as test data for experiments. The corresponding classification error will be obtained for each trial, and the mean of 10 results will be used as an estimate of the accuracy of the algorithm.

The variables were sorted according to their importance to age regression, and the combination of variables that did not significantly change the ability of the random forest regression model to distinguish age as the variables decreased was used as the final age-related microbial marker. Among them, markers derived from dental plaque include Prevotella loescheii, Kingella denitrificans, Leptotrichia BU064, Fusobacterium polymorpha subsp. nucleatum subsp.polymorphum), Bergeyella 602D02 (Bergeyella602D02), Cardiobacterium valvarum, Streptococcus mitis/Streptococcus pneumonia/Streptococcus infantis/Streptococcus oralis , Neisseria flava/Neisseria mucosa/Neisseria pharyngis, Campylobacter gracilis, Neisseria flavescens, 15 markers from saliva Species including Porphyromonas CW034 (Porphyromonas CW034), Streptococcus gordonii (Streptococcus gordonii), Veillonella atypical/Veillonella dispar/Veillonella parvula, oral digestion Streptococcus (Peptostreptococcus stomatis), Streptococcus parasanguinis/Streptococcus oralis, Leptotrichia BU064, (Porphyromonas catoniae), TM7 oral taxon 352, Prevotella Oral taxon 299 (Prevotella oral taxon 299), Prevotella melaninogenica (Prevotella melaninogenica), Eubacterium sulci/Eubacterium infirmum, Bergeyella 602D02 (Bergeyella 602D02), Neisseria aureus ( Neisseria flavescens), Neisseria meningitidis/Neisseria polysaccharide (Neisseria seria meningitides/Neisseria polysaccharea), Granulicatella elegans (Figure 3).

Using the random forest method, age-related microbial markers are used as input variables, and the actual month age corresponding to each sample is used as the sample information, and it is regressed to the discrete output variable (predicted month age), and finally an optimized test child is established. Models of the biological age of individuals.

(9) Application and performance of the optimized model (Figure 4)

The optimized model is applied to different groups, that is, the composition and abundance of age-related microbial markers in each sample are used as input variables, and the age detection model is used for regression analysis to obtain the biological age of the sample at this time. The results show that : In the healthy group, it can be seen that the biological age obtained by the oral microbial community is basically consistent with the chronological age; in the caries group, the biological age obtained by the oral microbial community is significantly lower than the chronological age (t test, p<0.05 ), suggesting that the occurrence of oral diseases potentially inhibits the maturity of the flora, leading to a decrease in the age of the oral flora. Individual oral health status in children.

The regression analysis method based on random forest of the present invention can refer to Breiman L (2001) Random forests.Mach Learn 45:5-32.) and (Knights D, Costello EK, Knight R.Supervised classification of human microbiota.FEMS Microbiol Rev.2011Mar;35(2):343-59.doi:10.1111/j.1574-6976.2010.00251.x.Epub 2010Oct 7.Review.PubMed PMID:21039646.

Of course, the above descriptions are not intended to limit the present invention, and the present invention is not limited to the above examples. Any changes, modifications, additions or substitutions made by those skilled in the art within the implementation scope of the present invention shall be Belong to the protection scope of the present invention.

Claims (7)

1. the method obtaining child's individuality biological age based on oral microbial community, it is characterised in that Comprise the following steps:

Data collection: collect child's individuality oral cavity sample of multiple time point；

Data convert: extract the DNA information obtaining oral cavity sample, utilize bioinformatics method by described DNA information is converted into oral microbial community information；

The Primary Construction of data model: using the oral microbial community information of acquisition as input variable, utilize Random forest method, returns it age information, and Primary Construction is based on oral microbial community information The preliminary mathematical model of detection biological age；

The optimization of mathematical model and determining: sort at the importance degree of model according to variable, do not affecting mould Under type performance premise, the combination of simplified model variable, finally determines the model that child's Individual Age detects；

The detection of child's individuality biological age: using desired microorganisms group information as input variable, has utilized The mathematical model set up carries out regression analysis, it is thus achieved that the child's individuality now biological age detected.

One the most according to claim 1 obtains child's individuality biological age based on oral microbial community Method, it is characterised in that described oral cavity sample is dental plaque sample on saliva or gum.

One the most according to claim 1 obtains child's individuality biological age based on oral microbial community Method, it is characterised in that the described oral microbial community information that is converted into by DNA information includes following step Rapid:

16s RNA or the full-length genome information of DNA information is obtained by high-flux sequence means；

Carry out horizontal bacterial strains information from door to kind for 16s RNA or full-length genome information to incorporate into；

Add up each sample sequence number of each species in kind of level, and the sequence obtained with this population of samples respectively Columns calculates its ratio, thus obtains the relative abundance of each each species.

One the most according to claim 1 obtains child's individuality biological age based on oral microbial community Method, it is characterised in that the Primary Construction of described data model, comprise the following steps:

1) composition and the relative abundance thereof of the whole thin species level of the oral microorganism obtained are become as input Amount；

2) utilize random forest method, the age information that child is individual is returned, tentatively by input variable Build preliminary mathematical model based on oral microbial community infomation detection biological age.

One the most according to claim 1 obtains child's individuality biological age based on oral microbial community Method, it is characterised in that the optimization of described data model and determining, comprise the following steps:

1) each variable importance journey to model performance of the kind representing bacterium in preliminary mathematical model is obtained Degree；

2) according to variable, model importance degree is sorted from small to large, gradually reduces variable quantity, utilize with Machine forest method, carries out the regression analysis to the age, it is thus achieved that the model of different variable combinations；

3) evaluate the combination of simplification variable under not reducing model performance premise, be defined as age correlated variables, So that it is determined that final optimization pass model.

One the most according to claim 1 obtains child's individuality biological age based on oral microbial community Method, it is characterised in that the detection of described child's individuality biological age, comprise the following steps:

1) DNA of child's individuality oral cavity sample is obtained；

2) utilize bioinformatics method that DNA information is converted to oral microbial community information；

3) relative abundance of the individual age correlated variables of child is obtained；

4) random forest method is utilized, using the composition of age correlated variables and abundance thereof as variable, to foundation Age detection model carry out regression analysis, it is thus achieved that the individual biological age now of child.

One the most according to claim 1 obtains child's individuality biological age based on oral microbial community Method, it is characterised in that also include: biological age individual for the child obtained is entered with its actual age Row contrast, knows the now growth promoter situation of child, if i.e. biological age is less than actual age, then carries Show that this child has owing to the factors such as disease cause the possibility of growth retardation；If biological age is equal to reality Age, then point out this upgrowth and development of children situation normal；If biological age is higher than actual age, then point out Child has the possibility of precocity.