Screening method and application of potential biomarkers of tuberculosis plasma

Abstract

The invention discloses a tuberculosis plasma potential biomarker screening method and application, and relates to the technical field of biomarker screening, the method comprises the following steps: collecting a plasma sample experimental group and a control group of tuberculosis, pretreating, and detecting and analyzing by adopting a liquid chromatography-mass spectrometry technology; dividing a training set and a verification set, and performing multivariate statistical analysis and univariate analysis; taking an intersection of the significant difference metabolite training set and the significant difference metabolite verification set to obtain an overlapped difference metabolite set; a linear regression algorithm, a decision tree algorithm and a gradient lifting algorithm are adopted to screen the overlapped difference metabolite set, and an intersection and a tuberculosis plasma potential biomarker are taken. The tuberculosis plasma potential biomarkers are screened by combining metabonomics, multivariate statistical analysis, univariate analysis, a linear regression algorithm, a decision tree algorithm and a gradient boosting algorithm, and the screening accuracy is improved.

Description

Technical Field

[0001] The present application relates to the technical field of biomarker screening, and in particular to a method and application for screening potential biomarkers in tuberculosis plasma. Background Art

[0002] Tuberculosis (TB) is a global infectious disease caused by Mycobacterium tuberculosis (M. tuberculosis), a persistent and significant threat to human health for millennia. Despite significant advances in diagnostic techniques, treatments, and prevention strategies, early and accurate TB diagnosis, disease progression monitoring, and effective treatment response assessment remain challenging. Currently, detecting biomarkers in plasma is a useful adjunct approach; therefore, research is needed to identify potential TB biomarkers in plasma.

[0003] In the prior art, Chinese patent CN114235990A discloses a method and application for screening serum markers for tuberculosis. The method involves subjecting serum from tuberculosis patients (experimental group) and serum from healthy individuals (control group) to biomimetic affinity or affinity chromatography to prepare biomimetic affinity or affinity chromatography antibody columns for the experimental group and control group, respectively. Culture filtrate from Mycobacterium tuberculosis or Mycobacterium tuberculosis cytoplasmic proteins are chromatographed on the biomimetic affinity or affinity chromatography antibody columns for the experimental and control groups, respectively. The Mycobacterium tuberculosis proteins bound to the serum antibodies from the experimental and control groups are analyzed to preliminarily obtain serum markers for tuberculosis detection. The preliminarily selected serum markers for tuberculosis detection are then further analyzed.

[0004] After Mycobacterium tuberculosis infects the human body, it triggers a series of complex immune responses in the host. This immune response not only involves the activation and regulation of immune system cells, but is also accompanied by changes in the metabolic state of the whole body, including but not limited to energy metabolism, lipid metabolism, amino acid metabolism, etc. These metabolic changes may show different characteristics in different stages of tuberculosis (such as latent period and active period), so there is the possibility of screening metabolites as biomarkers of tuberculosis. However, the above-mentioned prior art only analyzes Mycobacterium tuberculosis proteins and does not consider the possibility of screening metabolites as biomarkers of tuberculosis. The screening accuracy of potential biomarkers of tuberculosis plasma is poor. Summary of the Invention

[0005] The present application provides a method and application for screening potential biomarkers of tuberculosis plasma to solve the problem that the existing technology does not consider the possibility of screening metabolites as biomarkers of tuberculosis and the screening accuracy of potential biomarkers of tuberculosis plasma is poor.

[0006] In one aspect, the present application provides a method for screening potential biomarkers of tuberculosis plasma, comprising the following steps:

[0007] Step 1: Collect plasma samples from the experimental group and the control group of tuberculosis, and after pretreatment, use liquid chromatography-mass spectrometry technology for detection and analysis to obtain the metabolite spectrum data set of the experimental group and the metabolite spectrum data set of the control group.

[0008] Step 2: The metabolite spectrum data set of the experimental group and the metabolite spectrum data set of the control group are divided into a training set and a validation set after one-to-one correspondence, and multivariate statistical analysis and univariate analysis are performed to obtain a training set of significantly different metabolites and a validation set of significantly different metabolites.

[0009] Step three: taking the intersection of the significantly different metabolite training set and the significantly different metabolite validation set to obtain an overlapping differential metabolite set.

[0010] Step 4: Linear regression algorithm, decision tree algorithm and gradient boosting algorithm are used to screen the overlapping differential metabolite sets and take the intersection to obtain core differential metabolites as potential biomarkers for tuberculosis plasma.

[0011] In one possible implementation, in step 1, the preprocessing includes:

[0012] Samples from the experimental group or control group were placed in a centrifuge tube, methanol-water solution was added, and the samples were quickly frozen in liquid nitrogen. After thawing on ice, they were vortexed, ultrasonicated, and centrifuged in sequence. The supernatant was placed in a new centrifuge tube, freeze-dried into dry powder, and dissolved in methanol-water solution.

[0013] In a possible implementation, in step 2, the multivariate statistical analysis includes principal component analysis and partial least squares discriminant analysis.

[0014] In a possible implementation, in step 2, the univariate analysis includes: testing the statistical significance of each metabolite between the experimental group and the control group based on the t-test method, and calculating the difference fold.

[0015] In a possible implementation, in step three, after obtaining the overlapping differential metabolite set, pathway enrichment analysis is performed on the overlapping differential metabolite set to obtain enriched pathways.

[0016] In a possible implementation, in step 4, the linear regression algorithm adopts a LASSO regression algorithm.

[0017] In a possible implementation, in step 4, the decision tree algorithm adopts a random forest algorithm.

[0018] In a possible implementation, in step 4, the gradient boosting algorithm adopts the XGBoost gradient boosting algorithm.

[0019] On the other hand, the present application provides the use of potential biomarkers in the preparation of diagnostic products for detecting tuberculosis, and the potential biomarkers are obtained by the tuberculosis plasma potential biomarker screening method of the present application.

[0020] In a possible implementation, the potential biomarkers include angiotensin IV and glycochenodeoxycholic acid.

[0021] The method and application of screening potential biomarkers of tuberculosis plasma in this application have the following advantages:

[0022] By combining metabolomics, multivariate statistical analysis, univariate analysis, linear regression algorithm, decision tree algorithm and gradient boosting algorithm to screen potential biomarkers in tuberculosis plasma, the screening accuracy of potential biomarkers in tuberculosis plasma was improved.

[0023] The sensitivity and specificity of the potential biomarkers angiotensin IV and glycochenodeoxycholic acid obtained by the method for screening potential biomarkers for tuberculosis plasma of the present application were evaluated. The area under the receiver operating characteristic curve of angiotensin IV was above 0.99, and the area under the receiver operating characteristic curve of glycochenodeoxycholic acid was above 0.71, indicating that angiotensin IV and glycochenodeoxycholic acid (especially angiotensin IV) have high predictive accuracy as potential biomarkers for tuberculosis plasma and can be used to prepare diagnostic products for detecting tuberculosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0025] Figure 1 A schematic diagram of the process for screening potential biomarkers of tuberculosis plasma provided in the embodiments of the present application;

[0026] Figure 2 This is a PCA diagram of metabolites in the negative ion mode of the training set provided in the embodiment of the present application;

[0027] Figure 3 This is a metabolite PCA diagram in the positive ion mode of the training set provided in the embodiment of the present application;

[0028] Figure 4 This is a PCA diagram of metabolites in the negative ion mode of the validation set provided in the examples of this application;

[0029] Figure 5This is a PCA diagram of metabolites in the positive ion mode of the validation set provided in the examples of this application;

[0030] Figure 6 Volcano plot of differential metabolites in negative ion mode of the significant differential metabolite training set provided in the embodiment of the present application;

[0031] Figure 7 Volcano plot of differential metabolites in the positive ion mode of the significant differential metabolite training set provided in the embodiment of the present application;

[0032] Figure 8 Volcano plot of differential metabolites in negative ion mode of the validation set of significant differential metabolites provided in the examples of this application;

[0033] Figure 9 Volcano plot of differential metabolites in the positive ion mode of the validation set of significant differential metabolites provided in the examples of this application;

[0034] Figure 10 A Venn diagram of overlapping differential metabolites in negative ion mode provided in an embodiment of the present application;

[0035] Figure 11 A Venn diagram of overlapping differential metabolites in positive ion mode provided in an embodiment of the present application;

[0036] Figure 12 A differential metabolite enrichment pathway diagram provided in the examples of this application;

[0037] Figure 13 This is a graph showing the results of the LASSO regression algorithm for screening differential metabolites provided in the examples of this application;

[0038] Figure 14 This is a graph showing the results of screening differential metabolites using the random forest algorithm provided in the examples of this application;

[0039] Figure 15 This is a graph showing the results of screening differential metabolites using the XGBoost gradient enhancement algorithm provided in the examples of this application;

[0040] Figure 16 This is a Venn diagram of the intersection of differential metabolites screened by the LASSO regression algorithm, random forest algorithm, and XGBoost gradient boosting algorithm provided in the embodiments of the present application;

[0041] Figure 17 Box plots of the seven core differential metabolites provided in the examples of this application;

[0042] Figure 18 Receiver operating characteristic curves for the experimental and control groups under the angiotensin IV training set provided in the examples of the present application;

[0043] Figure 19 This is a graph of the receiver operating characteristic curves of the experimental group and the control group under the training set of glycochenodeoxycholic acid provided in the examples of the present application;

[0044] Figure 20 Receiver operating characteristic curves for the experimental and control groups under the angiotensin IV validation set provided in the examples of the present application;

[0045] Figure 21 This is a graph of the receiver operating characteristic curves for the experimental and control groups under the validation set of glycochenodeoxycholic acid provided in the examples of the present application;

[0046] Figure 22 This is a correlation diagram of the seven core differential metabolites provided in the examples of this application;

[0047] Figure 23 Correlation analysis diagram of all identified metabolites in the experimental group and control group provided in the examples of this application;

[0048] Figure 24 This is a correlation analysis diagram of angiotensin IV provided in the examples of this application. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0050] like Figure 1 As shown, the present invention provides a method for screening potential biomarkers of tuberculosis plasma, comprising the following steps:

[0051] Step 1: Collect plasma samples from the experimental group and the control group of tuberculosis, and after pretreatment, use liquid chromatography-mass spectrometry technology for detection and analysis to obtain the metabolite spectrum data set of the experimental group and the metabolite spectrum data set of the control group.

[0052] Step 2: The metabolite spectrum data set of the experimental group and the metabolite spectrum data set of the control group are divided into a training set and a validation set after one-to-one correspondence, and multivariate statistical analysis and univariate analysis are performed to obtain a training set of significantly different metabolites and a validation set of significantly different metabolites.

[0053] Step three: taking the intersection of the significantly different metabolite training set and the significantly different metabolite validation set to obtain an overlapping differential metabolite set.

[0054] Step 4: Linear regression algorithm, decision tree algorithm and gradient boosting algorithm are used to screen the overlapping differential metabolite sets and take the intersection to obtain core differential metabolites as potential biomarkers for tuberculosis plasma.

[0055] Specifically, in this example, plasma samples from patients with pulmonary tuberculosis aged 18 to 60 years, recruited from the Xinjiang Uyghur Autonomous Region Infectious Diseases Hospital between January and March 2024, were used as the experimental group (TB group). These patients were diagnosed according to the Chinese diagnostic guidelines (Sun et al., 2023). Inclusion criteria included: 1) patients with positive etiology and pathology results; 2) patients with at least one symptom or sign associated with tuberculosis; 3) patients with imaging evidence consistent with tuberculosis; and 4) patients with a positive tuberculin skin test or interferon-γ release assay. Exclusion criteria included: 1) patients with comorbidities such as diabetes, autoimmune diseases, lung tumors, or other lung infections; and 2) patients who had received immunomodulators or anti-tuberculosis treatment. Healthy volunteers who underwent health examinations at the Xinjiang Uyghur Autonomous Region Children's Hospital were recruited as the control group (HC group). All healthy volunteers were confirmed to be free of Mycobacterium tuberculosis (MTB) infection, lung disease, immunodeficiency, or immunosuppression.

[0056] Exemplarily, in step 1, the preprocessing includes:

[0057] Samples from the experimental group or control group were placed in a centrifuge tube, methanol-water solution was added, and the samples were quickly frozen in liquid nitrogen. After thawing on ice, they were vortexed, ultrasonicated, and centrifuged in sequence. The supernatant was placed in a new centrifuge tube, freeze-dried into dry powder, and dissolved in methanol-water solution.

[0058] Specifically, in this embodiment, the specific implementation of step one includes: taking 100 μL of the sample from the experimental group or the control group and placing it in a centrifuge tube, adding 400 μL of 80% methanol aqueous solution, placing it in liquid nitrogen for quick freezing for 5 minutes, thawing on ice and vortexing for 30 seconds, sequentially performing ultrasonic treatment for 6 minutes and 5000 rpm, 4°C centrifugation for 1 minute, taking the supernatant into a new centrifuge tube, freeze-drying it into a dry powder, adding the corresponding 10% methanol aqueous solution to dissolve it according to the volume of the supernatant taken, using liquid chromatography-mass spectrometry for detection and analysis, and adding the obtained experimental group metabolite spectrum data to the experimental group metabolite spectrum data set, and the control group metabolite spectrum data to the control group metabolite spectrum data set. In other possible embodiments, the preprocessing parameters can also be set to other possible values.

[0059] Exemplarily, in step 2, the multivariate statistical analysis includes principal component analysis and partial least squares discriminant analysis.

[0060] Specifically, in this embodiment, the metabolite spectrum data set of the experimental group and the metabolite spectrum data set of the control group were matched one by one and then converted using the metabolomics data processing software metaX. Then, they were divided into a training set and a validation set. The training set corresponded to the data of 72 active tuberculosis patients and 78 healthy controls, and the validation set corresponded to the data of 30 tuberculosis patients and 30 healthy controls. Principal component analysis of the training set and validation set can help identify differences between samples and potential chemical changes. In this embodiment, four PCA graphs were generated by principal component analysis, as shown in Figure 2. Figures 2 to 5 As shown, including: metabolite PCA graph in negative ion mode of training set, metabolite PCA graph in positive ion mode of training set, metabolite PCA graph in negative ion mode of validation set, and metabolite PCA graph in positive ion mode of validation set. The similarities and differences between samples can be visualized through the four PCA graphs. Figures 2 to 5 The tuberculosis group in the experiment group.

[0061] Specifically, partial least squares discriminant analysis is a supervised pattern recognition method that uses partial least squares regression to establish a relationship model between metabolite expression and sample category, thereby predicting the sample category. In this embodiment, partial least squares discriminant analysis is performed on the training set and the validation set to obtain the contribution index (VIP value) of each metabolite.

[0062] Illustratively, in step 2, the univariate analysis includes: testing the statistical significance of each metabolite between the experimental group and the control group based on the t-test method, and calculating the difference fold.

[0063] Specifically, the process of testing the statistical significance of each metabolite between the experimental group and the control group based on the t-test method is as follows:

[0064] Determine the hypothesis: Null hypothesis, there is no significant difference in the expression levels of metabolites between the two groups; Alternative hypothesis, there is a significant difference in the expression levels of metabolites between the two groups.

[0065] Calculate the t statistic: Calculate the mean and standard deviation of the two groups of metabolites and use the t-test formula to calculate the t statistic.

[0066] Find the statistical significance indicator (P value): Based on the t-statistic and the degrees of freedom (sample size - 1), find the corresponding P value in the t-distribution table. You can also use statistical software (such as SPSS, R, etc.) to directly calculate the P value.

[0067] Specifically, the fold difference (FC value) is equal to the ratio of the expression levels of each metabolite between the experimental group and the control group.

[0068] Specifically, in this example, differential metabolites were screened using VIP value, P value, and FC value, with the screening criteria being: VIP value greater than 1, P value less than 0.05, and log2FC value greater than 1. In this example, 282 differential metabolites were screened from the training set as the significantly differential metabolite training set, and 297 differential metabolites were screened from the validation set as the significantly differential metabolite validation set.

[0069] Specifically, in this embodiment, four differential metabolite volcano plots were drawn using the ggplot2 toolkit in R software, as shown in FIG. Figures 6 to 9 As shown, including: the volcano plot of differential metabolites in the negative ion mode of the significant differential metabolite training set, the volcano plot of differential metabolites in the positive ion mode of the significant differential metabolite training set, the volcano plot of differential metabolites in the negative ion mode of the significant differential metabolite validation set, and the volcano plot of differential metabolites in the positive ion mode of the significant differential metabolite validation set. Figures 6 to 9 It can be seen that 35 differential metabolites were downregulated and 67 differential metabolites were upregulated in the negative ion mode of the significant differential metabolite training set; 84 differential metabolites were downregulated and 96 differential metabolites were upregulated in the positive ion mode of the significant differential metabolite training set; 31 differential metabolites were downregulated and 75 differential metabolites were upregulated in the negative ion mode of the significant differential metabolite validation set; 74 differential metabolites were downregulated and 117 metabolites were upregulated in the positive ion mode of the validation set.

[0070] Specifically, in this embodiment, in step 3, the intersection of the 282 differential metabolites in the significant differential metabolite training set and the 297 differential metabolites in the significant differential metabolite validation set was taken to obtain 214 overlapping differential metabolites as the overlapping differential metabolite set, including 72 overlapping differential metabolites in the significant differential metabolite training set and the significant differential metabolite validation set in the negative ion mode and 142 overlapping differential metabolites in the positive ion mode. Figure 10 and Figure 11 Shown are the Venn diagrams of overlapping differential metabolites in negative ion mode and positive ion mode, respectively. Figure 10 and Figure 11 The anions in it are negative ions, and the cations are positive ions.

[0071] Illustratively, in step three, after obtaining the overlapping differential metabolite set, pathway enrichment analysis is performed on the overlapping differential metabolite set to obtain enriched pathways.

[0072] Specifically, in this example, in order to explore the biological functions of 214 overlapping differential metabolites, the KEGG bioinformatics database was used to perform pathway enrichment analysis on the overlapping differential metabolite set, such as Figure 12Shown is a map of differential metabolite enrichment pathways. Differential metabolites were primarily enriched in five major categories: cellular processes, environmental information processing, human diseases, metabolism, and body systems. Of the 214 overlapping differential metabolites, 157 were primarily enriched in 25 pathways, including all metabolic pathways, lipid metabolism, and amino acid metabolism.

[0073] Exemplarily, in step 4, the linear regression algorithm adopts the LASSO regression algorithm.

[0074] Exemplarily, in step 4, the decision tree algorithm adopts a random forest algorithm.

[0075] Exemplarily, in step 4, the gradient boosting algorithm adopts the XGBoost gradient boosting algorithm.

[0076] Specifically, in this embodiment, the LASSO package, randomForest package, and XGBoost package in R language are used to implement the LASSO regression algorithm, random forest algorithm, and XGBoost gradient boosting algorithm. The three algorithms are used to screen the overlapping differential metabolite sets, such as Figures 13 to 15 As shown in Figure 2, the LASSO regression algorithm screened out 10 differential metabolites, the random forest algorithm screened out 17 differential metabolites, and the XGBoost gradient enhancement algorithm screened out 7 differential metabolites. Figure 16 As shown in the figure, the intersection of the differential metabolites screened by the LASSO regression algorithm, random forest algorithm (randomForest), and XGBoost gradient enhancement algorithm was taken to obtain 7 core differential metabolites: angiotensin IV, glycochenodeoxycholic acid, methylindole-3-acetic acid, sorbitol, asparagine-phenylalanine, benzamide, and carbadox.

[0077] Among them, the two final core differential metabolites selected by all three algorithms were angiotensin IV and glycochenodeoxycholic acid.

[0078] The embodiments of the present application also provide the use of potential biomarkers in the preparation of diagnostic products for detecting tuberculosis. The potential biomarkers are obtained by the method for screening potential biomarkers in tuberculosis plasma of the present application.

[0079] Exemplarily, the potential biomarkers include angiotensin IV and glycochenodeoxycholic acid.

[0080] In a possible embodiment, in order to graphically represent the measurement statistics of the central tendency and the measurement statistics of the dispersion, and to be able to see the changes and differences of the differential metabolites in different experimental groups, a box plot is drawn for the seven core differential metabolites screened, such as Figure 17 As shown. Figure 17It can be seen that the seven core differential metabolites were significantly different between the experimental group (TB group) and the control group (HC group).

[0081] like Figures 18 to 21 As shown in the figure, the area under the receiver operating characteristic curve (ROC curve) (AUC value) was used to evaluate the sensitivity and specificity of angiotensin IV and glycochenodeoxycholic acid for the diagnosis of tuberculosis. The areas under the ROC curve of the training set and validation set of angiotensin IV were 0.9990 and 0.9911, respectively, and the areas under the ROC curve of the training set and validation set of glycochenodeoxycholic acid were 0.7128 and 0.7345, respectively, indicating that angiotensin IV and glycochenodeoxycholic acid (especially angiotensin IV) have high predictive accuracy as potential biomarkers for tuberculosis in plasma.

[0082] like Figure 22 As shown in the table, the values in the table are r values. The positive and negative signs represent positive and negative correlations. The closer the absolute value of the r value is to 1, the stronger the correlation. The * in the table represents high significance. All significance indicators (P values) in the table are less than 0.05, indicating high significance. It can be seen that the seven core differential metabolites have significant correlations.

[0083] In one possible embodiment, in order to reveal the relationship between differential metabolites, the pathology-related lipid modules of the experimental group (TB) patients were explored relative to the control group (HC). Figure 23 As shown, only the correlation analysis of all identified metabolites with a P value less than 0.05 in the experimental and control groups is shown; Figure 24 As shown, it is from Figure 23 Correlation analysis of angiotensin IV was extracted.

[0084] The embodiments of the present application improve the accuracy of screening potential biomarkers for tuberculosis plasma by combining metabolomics, multivariate statistical analysis, univariate analysis, linear regression algorithm, decision tree algorithm and gradient boosting algorithm to screen potential biomarkers for tuberculosis plasma.

[0085] The sensitivity and specificity of the potential biomarkers angiotensin IV and glycochenodeoxycholic acid obtained by the method for screening potential biomarkers for tuberculosis plasma of the present application were evaluated. The area under the receiver operating characteristic curve of angiotensin IV was above 0.99, and the area under the receiver operating characteristic curve of glycochenodeoxycholic acid was above 0.71, indicating that angiotensin IV and glycochenodeoxycholic acid (especially angiotensin IV) have high predictive accuracy as potential biomarkers for tuberculosis plasma and can be used to prepare diagnostic products for detecting tuberculosis.

[0086] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0087] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A method for screening potential biomarkers of tuberculosis plasma, characterized in that: The following steps are involved: Step 1: Collect plasma samples from the experimental and control groups of tuberculosis patients, and after pretreatment, use liquid chromatography-mass spectrometry to detect and analyze them to obtain metabolite spectrum data sets of the experimental group and the control group; Step 2: The metabolite profile dataset of the experimental group and the metabolite profile dataset of the control group are divided into a training set and a validation set after one-to-one correspondence, and multivariate statistical analysis and univariate analysis are performed to obtain a training set of significantly different metabolites and a validation set of significantly different metabolites; Step 3, taking the intersection of the significantly different metabolite training set and the significantly different metabolite validation set to obtain an overlapping differential metabolite set; Step 4: Linear regression algorithm, decision tree algorithm and gradient boosting algorithm are used to screen the overlapping differential metabolite sets and take the intersection to obtain core differential metabolites as potential biomarkers for tuberculosis plasma.

2. The method for screening potential biomarkers of tuberculosis plasma according to claim 1, characterized in that: In step 1, the pretreatment includes: Samples from the experimental group or control group were placed in a centrifuge tube, methanol-water solution was added, and the samples were quickly frozen in liquid nitrogen. After thawing on ice, they were vortexed, ultrasonicated, and centrifuged in sequence. The supernatant was placed in a new centrifuge tube, freeze-dried into dry powder, and dissolved in methanol-water solution.

3. The method for screening potential biomarkers of tuberculosis plasma according to claim 1, characterized in that: In step 2, the multivariate statistical analysis includes principal component analysis and partial least squares discriminant analysis.

4. The method for screening potential biomarkers of tuberculosis plasma according to claim 1, characterized in that: In step 2, the univariate analysis includes: testing the statistical significance of each metabolite between the experimental group and the control group based on the t-test method, and calculating the difference fold.

5. The method for screening potential biomarkers of tuberculosis plasma according to claim 1, characterized in that: In step three, after obtaining the overlapping differential metabolite set, pathway enrichment analysis is performed on the overlapping differential metabolite set to obtain enriched pathways.

6. The method for screening potential biomarkers of tuberculosis plasma according to claim 1, characterized in that: In step 4, the linear regression algorithm adopts the LASSO regression algorithm.

7. The method for screening potential biomarkers of tuberculosis plasma according to claim 1, characterized in that: In step 4, the decision tree algorithm adopts the random forest algorithm.

8. The method for screening potential biomarkers of tuberculosis plasma according to claim 1, characterized in that: In step 4, the gradient boosting algorithm adopts the XGBoost gradient boosting algorithm.

9. Use of a potential biomarker in the preparation of a diagnostic product for detecting tuberculosis, characterized in that: The potential biomarker is obtained by the method for screening potential biomarkers of tuberculosis plasma according to any one of claims 1 to 8.

10. Use of the potential biomarker according to claim 9 in preparing a diagnostic product for detecting tuberculosis, characterized in that: The potential biomarkers include angiotensin IV and glycochenodeoxycholic acid.

Images