CN117747093B

CN117747093B - Method for constructing idiopathic pulmonary fibrosis diagnosis model and diagnosis system

Info

Publication number: CN117747093B
Application number: CN202410189821.8A
Authority: CN
Inventors: 酒连娣; 吕彬彬; 郭栋梁
Original assignee: Digital Health China Technologies Co Ltd
Current assignee: Digital Health China Technologies Co Ltd
Priority date: 2024-02-20
Filing date: 2024-02-20
Publication date: 2024-06-07
Anticipated expiration: 2044-02-20
Also published as: CN117747093A

Abstract

The invention relates to the technical field of genes, in particular to a construction method and a diagnosis system of an idiopathic pulmonary fibrosis diagnosis model; the system comprises a data acquisition module, a differential gene screening module, a characteristic gene screening module, a regression coefficient calculation module, a diagnosis model construction module and a diagnosis score calculation module, wherein the data acquisition module is used for acquiring gene expression profile chip data of an IPF patient through a GEO database, constructing a chip data training set, the differential gene screening module is used for screening differential genes through Bayesian inspection by utilizing the chip data training set, the characteristic gene screening module is used for screening characteristic genes based on a random forest classifier, the regression coefficient calculation module is used for fitting a logistic regression model in the training set based on the characteristic genes to obtain regression coefficients of the characteristic genes, the diagnosis model construction module is used for constructing a idiopathic pulmonary fibrosis diagnosis model, and the diagnosis module is used for calculating a diagnosis score through the idiopathic pulmonary fibrosis diagnosis model based on the expression quantity of the characteristic genes of a person to be detected; realizes rapid screening of idiopathic pulmonary fibrosis, realizes earlier, more accurate and simpler diagnosis of IPF, and improves prognosis.

Description

Method for constructing idiopathic pulmonary fibrosis diagnosis model and diagnosis system

Technical Field

The invention relates to the technical field of genes, in particular to a construction method and a diagnosis system of an idiopathic pulmonary fibrosis diagnosis model.

Background

Pulmonary fibrosis is the terminal change of a large class of pulmonary diseases characterized by proliferation of fibroblasts and aggregation of a large number of extracellular matrices with concomitant inflammatory injury and destruction of tissue architecture, i.e., abnormal repair of normal alveolar tissue after it has been damaged, leading to structural abnormalities (scar formation). The etiology of most patients with pulmonary fibrosis is unknown (idiopathic), and this group of diseases is called idiopathic interstitial pneumonia (Idiopathic Interstitial Pneumonia), which is a major category of interstitial pneumonia. The most common type of disease with pulmonary fibrosis lesions as the main manifestation in Idiopathic Interstitial Pneumonia (IIP) is idiopathic pulmonary fibrosis (Idiopathic Pulmonary Fibrosis), which is a serious interstitial lung disease that can lead to progressive loss of pulmonary function. Pulmonary fibrosis severely affects the respiratory function of the human body, manifesting as dry cough, progressive dyspnea (inadequate consciousness), and with increased disease and lung injury, patient respiratory function is continually worsened. The incidence and mortality of idiopathic pulmonary fibrosis increases year by year, with an average survival of only 2.8 years after diagnosis, and mortality higher than most tumors, known as a "neoplastic disease".

IPF diagnosis requires High Resolution CT (HRCT) and lung biopsy is required for some cases. IPF is often ignored at the time of initial diagnosis because it is clinically very similar to other diseases such as bronchitis, asthma and heart failure. Most patients have reached mid-to-late stage disease at the time of diagnosis, even though the treatment is worsening.

Therefore, it is highly desirable to build a diagnostic model that facilitates earlier, more accurate, simpler diagnosis of IPF, improving prognosis.

Disclosure of Invention

Aiming at the defects of the prior art, the invention aims to provide a refining method of vulcanized isobutene, which solves the problems of high corrosion level of copper and odor of products by adopting high corrosion components in the existing preparation method.

In order to solve the problems, the invention adopts the following technical scheme:

The idiopathic pulmonary fibrosis diagnosis system comprises a data acquisition module, a differential gene screening module, a characteristic gene screening module, a regression coefficient calculation module, a diagnosis model construction module and a diagnosis module;

The data acquisition module is used for acquiring the gene expression profile chip data of the IPF patient through the GEO database and constructing a chip data training set;

The differential gene screening module is used for analyzing differential expression genes of the IPF and control groups by using a chip data training set through Bayesian test, wherein the screening condition is p.adj <0.05 &| logFC | >0.5, and screening out differential genes;

the characteristic gene screening module is used for screening the differential genes based on a random forest classifier to obtain characteristic genes;

The regression coefficient calculation module is used for fitting a logistic regression model in a training set based on the characteristic genes to obtain regression coefficients of the characteristic genes;

the diagnosis model construction module is used for constructing an idiopathic pulmonary fibrosis diagnosis model according to the expression quantity of the characteristic genes and the regression coefficient thereof;

the diagnosis module is used for calculating a diagnosis score through the idiopathic pulmonary fibrosis diagnosis model based on the expression quantity of the characteristic genes of the person to be detected.

As an implementation manner, the obtaining the IPF patient gene expression profile chip data through the GEO database, and constructing the chip data training set includes:

Acquiring gene expression profile chip data of an IPF patient from a GEO database, wherein the gene expression profile chip data of the IPF patient comprises GSE132607, GSE38958 and GSE28221; the gene expression profile chip data of the IPF patient are subjected to log2 conversion, probes of the IPF patient are annotated, GSE132607 and GSE38958 are combined, the batch effect of the combined data is removed by using a removeBatchEffect function in R package limma, and a chip data training set is obtained through integration, and GSE28221 is used as a verification set.

As an embodiment, the screening the differential genes based on a random forest classifier includes:

The variable of the random forest classifier is set to be 18, the number of trees for calculating error rate is 1000, and characteristic genes with importance larger than 1 are screened out, wherein the characteristic genes comprise TLR10、GZMK、CD79A、NOG、P2RY10、KLRB1、N6AMT1、EIF1AX、GCNT4、FCRLA、CD40LG、CD69、ABCA13、RNASE3、CEACAM6、USP9Y、OLFM4、BPI、UTY、RPS4Y1 and DDX3Y.

As an embodiment, the regression coefficients of the characteristic genes: TLR10 is-1.34326, GZMK is 0.30552, CD79A is-1.1445, NOG is-0.54664, P2RY10 is 0.74687, KLRB1 is-1.56761, N6AMT1 is-0.83828, EIF1AX is-0.51833, GCNT4 is-0.27993, FCRLA is 1.70369, CD40LG is-2.0675, CD69 is 0.97147, ABCA13 is 1.54353, RNASE3 is-0.08833, CEACAM6 is 0.48974, USP9Y is 0.97489, OLFM4 is-0.49965, BPI is 0.39967, UTY is 1.07367, RPS4Y1 is-1.06128 and DDX3Y is 0.07629.

As an embodiment, the diagnostic model of idiopathic pulmonary fibrosis calculates a diagnostic score by:

1/(1+exp (-z), where ,z =[(-1.34326×TLR10)+(0.30552×GZMK)+(-1.1445×CD79A)+(-0.54664×NOG)+(0.74687×P2RY10)+(-1.56761×KLRB1)+(-0.83828×N6AMT1)+(-0.51833×EIF1AX)+(-0.27993×GCNT4)+(1.70369×FCRLA)+(-2.0675×CD40LG)+(0.97147×CD69)+(1.54353×ABCA13)+(-0.08833×RNASE3)+(0.48974×CEACAM6)+(0.97489×USP9Y)+(-0.49965×OLFM4)+(0.39967×BPI)+(1.07367×UTY)+(-1.06128×RPS4Y1)+(0.07629×DDX3Y)+32.73267]/10.

A method for constructing a diagnostic model of idiopathic pulmonary fibrosis comprises

Acquiring gene expression profile chip data of an IPF patient through a GEO database, and constructing a chip data training set;

Using a chip data training set to analyze differential expression genes of IPF and control groups through Bayesian test, wherein the screening condition is p.adj <0.05 +| logFC | >0.5, and screening out differential genes;

Screening out characteristic genes of the differential genes based on a random forest classifier;

fitting a logistic regression model in a training set based on the characteristic genes to obtain regression coefficients of the characteristic genes;

and constructing a idiopathic pulmonary fibrosis diagnosis model according to the expression quantity of the characteristic genes and the regression coefficient thereof.

1/(1+exp (-z)), where ,z =[(-1.34326×TLR10)+(0.30552×GZMK)+(-1.1445×CD79A)+(-0.54664×NOG)+(0.74687×P2RY10)+(-1.56761×KLRB1)+(-0.83828×N6AMT1)+(-0.51833×EIF1AX)+(-0.27993×GCNT4)+(1.70369×FCRLA)+(-2.0675×CD40LG)+(0.97147×CD69)+(1.54353×ABCA13)+(-0.08833×RNASE3)+(0.48974×CEACAM6)+(0.97489×USP9Y)+(-0.49965×OLFM4)+(0.39967×BPI)+(1.07367×UTY)+(-1.06128×RPS4Y1)+(0.07629×DDX3Y)+32.73267]/10.

The invention has the beneficial effects that: through the diagnosis model or the diagnosis system constructed by the invention, the characteristic genes and the regression coefficients thereof are obtained by screening the difference genes of the IPF and then aiming at the difference genes through a random forest classifier, and the IPF calculation is carried out through the constructed diagnosis model, so that the rapid screening of idiopathic pulmonary fibrosis is realized, the diagnosis of the IPF is carried out earlier, more accurately and more simply, and the prognosis is improved.

Drawings

FIG. 1 is a schematic diagram of an exemplary system for diagnosing idiopathic pulmonary fibrosis.

FIG. 2 is a graph of differential expression volcanic according to an embodiment of the present invention.

FIG. 3 is a thermal map of the expression level of 38 differential genes in a sample according to the embodiment of the present invention.

FIG. 4 is a scatter plot of 38 differential genes in an example of the present invention.

FIG. 5 is a graph showing the number of decision trees selected and the error rate according to an embodiment of the present invention.

FIG. 6 is a view showing the importance of genes in the examples of the present invention.

Fig. 7 is a ROC curve of a training set in an embodiment of the present invention.

Fig. 8 is a ROC curve of a validation set in an embodiment of the present invention.

FIG. 9 is a flowchart of a method for constructing a diagnostic model of idiopathic pulmonary fibrosis in an embodiment of the invention.

Detailed Description

The present invention will be described in further detail with reference to specific examples.

It should be noted that these examples are only for illustrating the present invention, and not for limiting the present invention, and simple modifications of the method under the premise of the inventive concept are all within the scope of the claimed invention.

Referring to fig. 1, a diagnosis system for idiopathic pulmonary fibrosis includes a data acquisition module 100, a differential gene screening module 200, a characteristic gene screening module 300, a regression coefficient calculation module 400, a diagnosis model construction module 500, and a diagnosis module 600;

the data acquisition module 100 is used for acquiring the gene expression profile chip data of the IPF patient through the GEO database and constructing a chip data training set.

The method comprises the steps of obtaining IPF patient gene expression profile chip data through a GEO database, and constructing a chip data training set, wherein the method comprises the following steps:

Obtaining gene expression profile chip data of an IPF patient from a GEO database, wherein the gene expression profile chip data of the IPF patient are shown in table 1 and comprise GSE132607, GSE38958 and GSE28221; the gene expression profile chip data of the IPF patient are subjected to log2 conversion, probes of the IPF patient are annotated, GSE132607 and GSE38958 are combined, the batch effect of the combined data is removed by using a removeBatchEffect function in R package limma, and a chip data training set is obtained through integration, and GSE28221 is used as a verification set.

TABLE 1 GEO database IPFmRNA expression profiling chip data

Data set	IPF	Normal state	Platform	Tissue of
					GSE132607	276	0	Gene Expression Array	PBMC
GSE38958	70	45	Gene Expression Array	PBMC
					GSE28221	120	19	Gene Expression Array	PBMC

The differential gene screening module 200 is used for analyzing differential expression genes of IPF and control groups by using a chip data training set through Bayesian test, screening conditions are p.adj <0.05 &| logFC | >0.5, and screening out differential genes. 38 differential genes were selected, see FIGS. 2 and 3.

The characteristic gene screening module 300 is used for screening the differential genes based on a random forest classifier.

Specifically, 38 difference genes of the chip data combined by GSE132607 and GSE38958 are put into a random forest classifier, in order to find the optimal variable number, random forest classification calculation is performed on all possible variable number cycles, the corresponding error rate is obtained, and finally 18 is selected as the optimal variable number, as shown in figure 4; and calculating the error rate of 1-2000 trees, wherein the error rate is not changed after the number of the trees reaches 1000, see fig. 5. Finally we select 18 variables, 1000 trees as the last calculated parameters. Subsequently we screened 21 signature genes with importance greater than 1, see figure 6.

The regression coefficient calculation module 400 is configured to fit a logistic regression model to the training set based on the characteristic genes, so as to obtain regression coefficients of the characteristic genes, as shown in table 2.

TABLE 2 regression coefficient table of characteristic genes

Characteristic gene	Regression coefficient	P value
			Intercept	32.73267	0.000102
TLR10	-1.34326	0.013703
			GZMK	0.30552	0.045683
CD79A	-1.1445	0.063288
			NOG	-0.54664	0.031332
P2RY10	0.74687	0.024423
			KLRB1	-1.56761	0.009573
N6AMT1	-0.83828	0.020989
			EIF1AX	-0.51833	0.046240
GCNT4	-0.27993	0.068536
			FCRLA	1.70369	0.028434
CD40LG	-2.0675	0.017964
			CD69	0.97147	0.033659
ABCA13	1.54353	0.042329
			RNASE3	-0.08833	0.065271
CEACAM6	0.48974	0.046443
			USP9Y	0.97489	0.025125
OLFM4	-0.49965	0.038818
			BPI	0.39967	0.046291
UTY	1.07367	0.014165
			RPS4Y1	-1.06128	0.05182
DDX3Y	0.07629	0.047289

The diagnostic model construction module 500 is used for constructing a idiopathic pulmonary fibrosis diagnostic model according to the expression level of the characteristic genes and regression coefficients thereof.

The diagnosis module 600 is used for calculating a diagnosis score by the idiopathic pulmonary fibrosis diagnosis model based on the expression level of the characteristic gene of the subject.

The diagnostic model of idiopathic pulmonary fibrosis calculates a diagnostic score by:

The predictive power of the ROC curve test model is used, and the AUC >0.7 is generally considered to be good for discrimination. The ROC curve of the training set is shown in fig. 7, the maximum sign index of the training set is 0.656, the area under the ROC curve AUC is 0.893 (95% ci 0.845-0.941), the optimal cut-off value is 0.875, and the sensitivity is 0.723 and the specificity is 0.933 under the cut-off value; the ROC curve part of the verification set is shown as 8, and the model has good prediction capability.

Referring to FIG. 9, a method for constructing a diagnostic model of idiopathic pulmonary fibrosis is provided, comprising

S100, acquiring gene expression profile chip data of an IPF patient through a GEO database, and constructing a chip data training set;

S200, analyzing differential expression genes of IPF and control groups by using a chip data training set through Bayesian test, wherein the screening condition is p.adj <0.05 &| logFC | >0.5, and screening out the differential genes;

s300, screening out characteristic genes from the differential genes based on a random forest classifier;

S400, fitting a logistic regression model in a training set based on the characteristic genes to obtain regression coefficients of the characteristic genes;

s500, constructing a idiopathic pulmonary fibrosis diagnosis model according to the expression quantity of the characteristic genes and the regression coefficients of the characteristic genes.

Wherein, screening the differential genes out characteristic genes based on a random forest classifier comprises the following steps:

Wherein, regression coefficients of each characteristic gene: TLR10 is-1.34326, GZMK is 0.30552, CD79A is-1.1445, NOG is-0.54664, P2RY10 is 0.74687, KLRB1 is-1.56761, N6AMT1 is-0.83828, EIF1AX is-0.51833, GCNT4 is-0.27993, FCRLA is 1.70369, CD40LG is-2.0675, CD69 is 0.97147, ABCA13 is 1.54353, RNASE3 is-0.08833, CEACAM6 is 0.48974, USP9Y is 0.97489, OLFM4 is-0.49965, BPI is 0.39967, UTY is 1.07367, RPS4Y1 is-1.06128 and DDX3Y is 0.07629.

Wherein the diagnostic model of idiopathic pulmonary fibrosis calculates a diagnostic score by:

Finally, it is noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be understood that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims

1. The idiopathic pulmonary fibrosis diagnosis system is characterized by comprising a data acquisition module, a differential gene screening module, a characteristic gene screening module, a regression coefficient calculation module, a diagnosis model construction module and a diagnosis module;

the diagnosis module is used for calculating a diagnosis score through the idiopathic pulmonary fibrosis diagnosis model based on the expression quantity of the characteristic genes of the person to be detected;

The method for obtaining the gene expression profile chip data of the IPF patient through the GEO database and constructing the chip data training set comprises the following steps:

acquiring gene expression profile chip data of an IPF patient from a GEO database, wherein the gene expression profile chip data of the IPF patient comprises GSE132607, GSE38958 and GSE28221; the method comprises the steps of performing log2 conversion on gene expression profile chip data of an IPF patient, annotating probes of the IPF patient, combining GSE132607 and GSE38958, removing batch effects of the combined data by using a removeBatchEffect function in R-packet limma, and integrating to obtain a chip data training set, wherein GSE28221 is used as a verification set;

Screening the differential genes based on a random forest classifier to obtain characteristic genes, wherein the screening comprises the following steps:

the variable of the random forest classifier is set to be 18, the number of trees for calculating error rate is 1000, and characteristic genes with importance larger than 1 are screened out, wherein the characteristic genes comprise TLR10、GZMK、CD79A、NOG、P2RY10、KLRB1、N6AMT1、EIF1AX、GCNT4、FCRLA、CD40LG、CD69、ABCA13、RNASE3、CEACAM6、USP9Y、OLFM4、BPI、UTY、RPS4Y1 and DDX3Y;

Regression coefficients of the characteristic genes: TLR10 is-1.34326, GZMK is 0.30552, CD79A is-1.1445, NOG is-0.54664, P2RY10 is 0.74687, KLRB1 is-1.56761, N6AMT1 is-0.83828, EIF1AX is-0.51833, GCNT4 is-0.27993, FCRLA is 1.70369, CD40LG is-2.0675, CD69 is 0.97147, ABCA13 is 1.54353, RNASE3 is-0.08833, CEACAM6 is 0.48974, USP9Y is 0.97489, OLFM4 is-0.49965, BPI is 0.39967, UTY is 1.07367, RPS4Y1 is-1.06128 and DDX3Y is 0.07629.

2. The idiopathic pulmonary fibrosis diagnostic system of claim 1, wherein the idiopathic pulmonary fibrosis diagnostic model calculates a diagnostic score by:

1/(1+exp(-z))，

wherein z= [ (-1.34326 x TLR 10) + (0.30552 x GZMK) +(-1.1445 x CD 79A) +(-0.5466)

4×NOG)+(0.74687×P2RY10)+(-1.56761×KLRB1)+(-0.83828×N6AMT1)+(-0.51833×EIF1AX)+(-0.27993×GCNT4)+(1.70369×FCRLA)+(-2.0675×CD40LG)+(0.97147×CD69)+(1.54353×ABCA13)+(-0.08833×RNASE3)+(0.48974×CEACAM6)+(0.97489×USP9Y)+(-0.49965×OLFM4)+(0.39967×BPI)+(1.07367×UTY)+(-1.06128×RPS4Y1)+(0.07629×DDX3Y)+32.73267]/10.

3. A method for constructing a diagnosis model of idiopathic pulmonary fibrosis is characterized by comprising the following steps of

Constructing a idiopathic pulmonary fibrosis diagnosis model according to the expression quantity of the characteristic genes and the regression coefficient of the idiopathic pulmonary fibrosis diagnosis model;

4. The method of constructing an idiopathic pulmonary fibrosis diagnostic model of claim 3, wherein the idiopathic pulmonary fibrosis diagnostic model calculates a diagnostic score by:

1/(1+exp(-z))；

wherein z= [ (-1.34326 x TLR 10) + (0.30552 x GZMK) +(-1.1445 x CD 79A) +(-0.54)

664×NOG)+(0.74687×P2RY10)+(-1.56761×KLRB1)+(-0.83828×N6AMT1)+(-0.51833×EIF1AX)+(-0.27993×GCNT4)+(1.70369×FCRLA)+(-2.0675×CD40LG)+(0.97147×CD69)+(1.54353×ABCA13)+(-0.08833×RNASE3)+(0.48974×CEACAM6)+(0.97489×USP9Y)+(-0.49965×OLFM4)+(0.39967×BPI)+(1.07367×UTY)+(-1.06128×RPS4Y1)+(0.07629×DDX3Y)+32.73267]/10.