Prognostic model of patient with colon adenocarcinoma combined with multiple adenomatous polyp and construction method of prognostic model

By combining metabolomics and proteomics analysis with the TCGA database, significantly differentially expressed metabolites and proteins were screened to construct a prognostic model for patients with colon adenocarcinoma complicated by multiple adenomatous polyps. This solved the problem that existing models failed to incorporate the evolution of colon adenoma to adenocarcinoma, achieving more efficient predictive results.

CN119049702APending Publication Date: 2024-11-29ZHONGSHAN HOSPITAL AFFILIATED TO FUDAN UNIV XIAMEN HOSPITAL
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410866318.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-01
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

Existing colorectal adenocarcinoma prediction models fail to effectively incorporate the colorectal adenoma-adenocarcinoma evolution process, resulting in poor predictive performance.

Method used

By combining metabolomics, proteomics, and database analysis, significantly differentially expressed metabolites and proteins were screened, co-expressed genes were identified, and univariate Cox proportional hazards regression analysis was performed to construct a prognostic model for patients with colorectal adenocarcinoma complicated with multiple adenomatous polyps.

Benefits of technology

It provides a more clinically relevant prognostic model that can efficiently and economically assess the prognosis and treatment options for patients with colon adenocarcinoma complicated by multiple adenomatous polyps, thus improving predictive accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119049702A_ABST
    Figure CN119049702A_ABST
Patent Text Reader

Abstract

The invention provides a prognosis model of a patient with colon adenocarcinoma combined with multiple adenoma polyp and a construction method thereof, and the construction method comprises the following steps: collecting a colon adenocarcinoma clinical sample and a multiple adenoma polyp clinical sample, respectively carrying out metabonomics detection to obtain metabolites, carrying out difference analysis on the metabolites, and screening out significant difference metabolites; carrying out proteomic analysis on the clinical sample, and screening out remarkably different proteins; after conjoint analysis is conducted on the remarkably differential metabolites and the remarkably differential proteins, differential genes are screened out on the basis of a TCGA-COAD database and a GTEx database, the remarkably differential proteins are converted into corresponding differential gene names through a Uniprot database, conjoint analysis is conducted on the remarkably differential protein genes and the differential genes, and co-expression genes are obtained; carrying out single-factor Cox proportional risk regression analysis by utilizing the co-expressed gene, and identifying a prognosis gene; and constructing a prognosis model according to the calculated multi-factor coefficient of the prognosis gene. The device is more clinically fit, and is convenient for clinicians to efficiently and economically use.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of prognostic models for colon adenocarcinoma, and in particular to a prognostic model for patients with colon adenocarcinoma combined with multiple adenomatous polyps and a method for constructing the model. Background Art

[0002] Colon adenoma polyp (COAP) is one of the most common diseases of the digestive tract and is considered a common precancerous disease of colon adenocarcinoma (COAD). It is currently believed that most colon adenocarcinomas evolve in a "colon adenoma-adenocarcinoma" manner. This pathway is known to be associated with many risk factors, including not only sociodemographics (such as age, gender, and race) and family history, but also lifestyle (such as smoking, alcohol history) and diet (such as large consumption of red meat and processed meat and low intake of dietary fiber). In addition, changes in the expression of tumor-related genes are also important genetic factors in the evolution of adenoma-adenocarcinoma. With the gradual popularization of colonoscopy, various endoscopic examinations, and the continuous improvement of AI-assisted endoscopic screening technologies, the detection rate of colon adenomas is increasing. Despite this, the incidence of colon cancer is still high.

[0003] Current treatments for colon cancer include surgery (primary and metastatic lesions), radiotherapy, and chemotherapy. However, in recent years, the 5-year survival rate of patients diagnosed with colon cancer has improved little. Therefore, there is an urgent need to find new biomarkers. The occurrence of colon cancer has the main pathways of classic adenoma-adenocarcinoma and hyperplastic polyps to serrated adenomas and then to adenocarcinomas. There are various changes in the relationship between cell apoptosis and proliferation at different stages. The embryological origins of the proximal and distal colon are also different, and there are also differences in morphology and physiological functions, showing the correlation and complexity between the morphological characteristics and biological characteristics of tumor cells. The occurrence and development of colon cancer is a multi-stage and multi-gene process. Oncogene activation, tumor suppressor gene inactivation, abnormal cell proliferation, and uncontrolled signal transduction pathways can all promote tumor occurrence. In recent years, bioinformatics statistical analysis has developed rapidly, and researchers can quickly analyze important pathogenic genes of colon cancer through public database data.

[0004] At present, some researchers have constructed a prediction model for immunotherapy of colon cancer, and screened differentially expressed immune-related genes through different statistical analyses; based on the screened immune-related genes and the proportional risk Cox regression model, multiple prognosis-related immune genes were obtained to predict the efficacy of immunotherapy for colon cancer patients, providing a new idea for clinicians. However, this method still has limitations. It only uses the sequencing results of foreign public databases or colon cancer tissue samples, without combining the evolution process of colon adenoma-adenocarcinoma, and the prediction model is not effective.

[0005] In view of this, the present application proposes a prognostic model for patients with colon adenocarcinoma combined with multiple adenomatous polyps and a method for constructing the model, which can more realistically approach the mechanism of colon adenoma-adenocarcinoma occurrence. Summary of the invention

[0006] In order to solve the problems that the existing model does not incorporate the evolution process of colon adenoma-adenocarcinoma and the prediction model effect is poor, the present application provides a prognosis model for patients with colon adenocarcinoma combined with multiple adenomatous polyps and a method for constructing it to solve the above technical defects.

[0007] According to a first aspect of the present invention, a method for constructing a prognostic model for patients with colon adenocarcinoma and multiple adenomatous polyps is proposed, comprising the following steps:

[0008] S1. Collect clinical samples of colon adenocarcinoma and multiple adenomatous polyps, perform metabolomics detection on them, obtain metabolites, perform differential analysis on the metabolites, and screen out significantly differential metabolites according to the set first threshold;

[0009] S2. Perform proteomic analysis on clinical samples of colon adenocarcinoma and multiple adenomatous polyps, and screen out significantly differentially expressed proteins based on the set second threshold;

[0010] S3. After joint analysis of significantly differential metabolites and significantly differential proteins, differential genes were screened out based on the TCGA-COAD database and the GTEx database, and significantly differential proteins were converted into differential protein genes. The significantly differential protein genes were jointly analyzed with differential genes to obtain co-expressed genes.

[0011] S4. Univariate Cox proportional hazards regression analysis was performed using co-expressed genes to identify prognostic genes;

[0012] S5. Construct a prognostic model based on the calculated multivariate coefficients of the prognostic genes.

[0013] In the above technical scheme, the construction method of the present invention utilizes TCGA and GTEx databases in conjunction with proteomics data (ZSXM-cohort) for joint analysis, and ultimately screens out prognostic genes, namely, colon cancer prognostic genes, which are more clinically relevant and convenient for clinicians to use efficiently and economically.

[0014] According to functional analysis (KEGG analysis), the pathways in which differentially expressed proteins in the proteome and differentially expressed metabolites in the metabolome are involved are compared, such as enriching metabolites and proteins that play the same functional pathway. The important purpose of the joint analysis of significantly differential metabolites and significantly differential proteins is to better study the protein expression regulation mechanism, form a more complete metabolic regulation network diagram, and be able to globally observe the expression correlation between proteins and metabolites at the multi-omics level. By projecting the differential proteins obtained by proteomics and the differential metabolites obtained by metabolomics to the KEGG pathway at the same time, the pathway data can be integrated intuitively and comprehensively. It can better screen out functional proteins that play an important role in the sequential evolution of colon adenomatous polyps to colon adenocarcinoma.

[0015] In a second aspect, the present application proposes a prognostic model for patients with colon adenocarcinoma and multiple adenomatous polyps, which is constructed using any of the construction methods described above.

[0016] Compared with the prior art, the beneficial results of the present invention are:

[0017] The present invention provides a prognosis prediction model for patients with colon adenocarcinoma combined with multiple adenomatous polyps based on proteomics (4D Label free), metabolomics (high-resolution Agilent1290Infinity LC-MS) and TCGA database joint analysis and a construction method thereof. The present invention deeply analyzes the immune cells, immune matrix, tumor mutations and IC50 drug sensitivity of the risk prognosis model, and finally screens out 4 characteristic genes (CSRP2, FABP4, TIMP1 and LRCH4), which are the prognosis genes for patients with colon adenocarcinoma combined with multiple adenomatous polyps, which are more in line with clinical practice and convenient for clinicians to use efficiently and economically. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0019] Figure 1 is a flow chart of a method for constructing a prognostic model for patients with colon adenocarcinoma and multiple adenomatous polyps according to the present application;

[0020] Figure 2 is a schematic diagram of metabolomics analysis;

[0021] Figure 3 is the PCA score plot for the positive ion mode;

[0022] Figure 4 is the PCA score plot for the positive ion mode;

[0023] Figure 5is the PLS-DA score graph in positive ion mode;

[0024] Figure 6 is the PLS-DA score graph in negative ion mode;

[0025] Figure 7 is the hierarchical clustering tree diagram of positive ion mode samples;

[0026] Figure 8 is the hierarchical clustering tree diagram of negative ion mode samples;

[0027] Fig. 9 This is the hierarchical clustering heat map of significantly different metabolites in positive ion mode;

[0028] Fig.10 It is the hierarchical clustering heat map of significant difference metabolites in negative ion mode;

[0029] Fig.11 is the positive ion mode chord diagram;

[0030] Fig.12 It is the negative ion mode chord diagram;

[0031] Fig.13 It is the heat map of differential metabolites clustering of KEGG pathway;

[0032] Fig.14 It is the KEGG enriched pathway bubble diagram;

[0033] Fig.15 is the KEGG enriched pathway bar graph;

[0034] Fig.16 is the differential abundance score map of all differential metabolic pathways;

[0035] Fig.17 It is the differential abundance score map of all differential metabolic pathways classified according to Pathway_Hierarchy;

[0036] Fig.18 is the correlation spectrum of the QC sample in positive ion mode;

[0037] Fig.19 is the correlation spectrum of the negative ion mode QC sample;

[0038] Fig. 20 is the Hotellings T2 plot of the overall sample in positive ion mode;

[0039] Fig.21 is the Hotellings T2 plot of the overall sample in negative ion mode;

[0040] Fig. 22 This is the MCC graph of the QC sample in positive ion mode;

[0041] Fig.23 This is the MCC graph of the negative ion mode QC sample;

[0042] Fig.24 is the relative standard deviation plot of the QC samples in positive ion mode;

[0043] Fig.25 is the relative standard deviation graph of the negative ion mode QC samples;

[0044] Fig.26 is a proteomic analysis diagram;

[0045] Fig. 27 is the cluster analysis diagram of differentially expressed proteins;

[0046] Fig.28 is the peptide ion mass deviation distribution diagram;

[0047] Fig.29 is the peptide ion score distribution plot;

[0048] Fig.30 It is the metabolomics and proteomics correlation analysis diagram;

[0049] Fig.31 This is the volcano plot analysis of differentially expressed genes in the TCGA-COAD database;

[0050] Fig.32 It is a Venn diagram analysis diagram of differentially expressed genes;

[0051] Fig.33 is the differential gene enrichment analysis diagram;

[0052] Fig.34 is the forest plot of differentially expressed genes;

[0053] Fig.35 is the cross-validation graph of differentially expressed genes;

[0054] Fig.36 It is the high-risk and low-risk dividing line of the prognostic model;

[0055] Fig.37 It is a heat map of high and low risk of model genes;

[0056] Fig.38 This is the ROC analysis chart of the prognostic model of TCGA data;

[0057] Fig.39 This is the ROC analysis diagram of the prognostic model of the GSE database;

[0058] Fig.40 This is the TCGA validation model diagram;

[0059] Fig.41It is the GSE validation model diagram;

[0060] Fig.42 is the expression and survival analysis graph of prognostic genes in colon adenocarcinoma;

[0061] Fig.43 This is the enrichment analysis diagram of the prognostic gene KEGG;

[0062] Fig.44 is the risk score versus matrix score, immune score, and ESTIMATE score plot;

[0063] Fig.45 is a heat map of the abundance of immune cell populations in high- and low-risk prognostic models;

[0064] Fig.46 It is a graph showing the correlation between the clinical characteristics of risk and seven immunomodulators and risk scores;

[0065] Fig.47 It is a waterfall plot of high-risk prognostic gene mutations;

[0066] Fig.48 It is a waterfall plot of low-risk prognostic gene mutations;

[0067] Fig.49 This is an analysis chart of drug sensitivity differences among patients with colon adenocarcinoma in different risk groups. DETAILED DESCRIPTION

[0068] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It should also be noted that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0069] The present invention uses clinical samples of patients with colon cancer and multiple adenomatous polyps of the colon for the first time, which more realistically presents the evolution mechanism of colon adenoma-colon cancer. The latest 4D proteomics and high-resolution Agilent 1290InfinityLC-MS metabolomics detection methods were used to screen significantly differentially expressed genes. These genes were further analyzed in conjunction with the TCGA and GTEx databases, and then a colon cancer prognosis model was constructed using univariate regression and Lass regression analysis to identify the prognosis of patients with different characteristics. In order to further verify the accuracy and effectiveness of the model, this application uses the GSE39582 database as an external validation to continue to evaluate the accuracy of the model.

[0070] At the same time, we explored the immune infiltrating cells, immune scores, stromal scores, tumor mutation load, immune checkpoint expression, and response to immunotherapy in patients with different characteristics. Finally, we screened out four characteristic genes (CSRP2, FABP4, TIMP1, and LRCH4), which are the prognostic genes for colon adenocarcinoma. These genes are more in line with clinical practice and are convenient for clinicians to use efficiently and economically, providing a theoretical basis for the comprehensive evaluation of the prognosis and treatment of patients with colon adenocarcinoma and multiple adenomatous polyps.

[0071] Figure 1 A flowchart of the method for constructing a prognostic model for patients with colon adenocarcinoma combined with multiple adenomatous polyps of the present application is shown, as Figure 1 As shown, the prognostic model construction method of the present application is:

[0072] 1. Clinical sample collection: 17 fresh tissues of colon adenocarcinoma (COAD) combined with multiple adenomatous polyps (COAP) were collected, including 6 females and 11 males. All of them were still in the direct family. Gastroenteroscopy showed no multiple colon polyps, and pathology confirmed colon adenocarcinoma and colon tubular adenoma.

[0073] 2. Metabolomics analysis: The samples were divided into two groups: colon adenocarcinoma and multiple colon adenomatous polyps. Each group was repeated 6 times, and metabolomics (high-resolution Agilent 1290Infinity LC-MS) was performed. A total of 1,838 metabolites were identified, including 1,125 positive ions and 713 negative ions; 64% of these metabolites were related to lipid molecules, and 14% were related to organic acids and their derivatives. According to the expression multiple FC>1.5 or expression multiple FC<0.67, OPLS-DA VIP>1 and statistical significance P value Pvalue<0.05, 276 significantly different metabolites were screened out.

[0074] Figure 2 A schematic diagram of metabolomics analysis is shown, where Figure 2 AB shows the anion and cation volcano plots of the metabolomics analysis. Based on univariate analysis, differential analysis was performed on all metabolites detected in the positive and negative ion modes (including unidentified metabolites). The differential metabolites with FC>1.5 or FC<0.67 and p value<0.05 were visualized in the form of a volcano plot. In order to avoid overfitting of metabolomics data during the modeling process, the permutation test was used to test the model to ensure the validity of the model. The PLS-DA model permutation test was performed. As the permutation retention gradually decreased, the R2 and Q2 of the random model gradually decreased, indicating that the original model did not have overfitting. The metabolomics detection and analysis model of the present invention has good robustness, indicating that the detected differential metabolites can truly reflect the characteristics of the metabolic expression spectrum of colon adenocarcinoma and colon adenoma tissues (such as Figure 2As shown in C and D, Figure 2 CD in the figure is the anion and cation PLS-DA diagram of metabolomics analysis).

[0075] PLS-DA permutation test: In order to avoid overfitting of the supervised model during the modeling process, the permutation test is used to test the model to ensure the effectiveness of the model. Figure 2 CD shows the permutation test diagram of the PLS-DA model of the example comparison group. As the permutation retention gradually decreases, the R2 and Q2 of the random model gradually decrease, indicating that the original model does not have overfitting and the model is robust. Figure 2 The horizontal axis in CD represents the permutation retention, that is, the proportion of the order of the Y variable consistent with the original model, and the vertical axis represents the values ​​of R2 and Q2. The green point represents R2, the blue point represents Q2, and the two dotted lines represent the regression lines of R2 and Q2 respectively. R2 and Q2 in the upper right corner indicate that the permutation retention is equal to 1, that is, the R2 and Q2 values ​​of the original model.

[0076] 3. Multi-dimensional analysis of metabolites:

[0077] 3.1 Principal Component Analysis (PCA)

[0078] Principal Component Analysis (PCA) is an unsupervised data analysis method that linearly recombines all the originally identified metabolites to form a new set of comprehensive variables. At the same time, it selects significantly different metabolites based on colon adenocarcinoma and colon multiple adenomatous polyps tissues so that these metabolites can reflect as much information as possible about the original variables, thereby achieving the purpose of dimensionality reduction. At the same time, principal component analysis of metabolites can also reflect the variability between and within sample groups in general. Therefore, in data analysis, the PCA method is generally used first to observe the overall distribution trend of samples between groups and the degree of difference between samples between groups. Perform PCA analysis on each comparison group separately. Taking the example comparison group as an example, the PCA score graph is shown as follows Figure 3 and Figure 4 shown. Figure 3 , Figure 4 In the figure, t[1] represents principal component 1, t[2] represents principal component 2, and the ellipse represents the 95% confidence interval. Points of the same color represent the biological replicates within the group, and the distribution of the points reflects the differences between and within the groups.

[0079] 3.1 Partial Least Squares Discriminant Analysis (PLS-DA)

[0080] Partial Least Squares Discrimination Analysis (PLS-DA) is a supervised discriminant analysis statistical method. This method uses partial least squares regression to establish a relationship model between metabolite expression and sample category to predict the sample category. Through the established discriminant model, differential metabolites related to grouping can be screened out from the data set (such as Figure 5 , Figure 6 It can be seen that the PLS-DA model can distinguish the two groups of samples. Figure 5 , Figure 6 In the figure, t[1] represents principal component 1, t[2] represents principal component 2, and the ellipse represents the 95% confidence interval. Points of the same color represent the biological replicates within the group, and the distribution of the points reflects the differences between and within the groups.

[0081] 4. Bioinformatics analysis of differential metabolites

[0082] Subsequent bioinformatics analysis was performed on the significantly different metabolites screened above (the metabolites must simultaneously meet OPLS-DA VIP>1 and p value<0.05, and have qualitative names), including cluster analysis, correlation analysis, pathway analysis, etc.

[0083] 4.1 Cluster analysis

[0084] In order to more comprehensively and intuitively display the relationship between samples and the differences in the expression patterns of metabolites in different samples, the expression levels of all samples and differential metabolites are subtracted from the average value of the group to which they belong and then divided by the root mean square of the group number for standardization. Then the distance matrix is ​​calculated and cluster analysis is performed using hierarchical clustering. Hierarchical clustering first needs to calculate the distance between samples or the distance between metabolites. Each time, the two clusters with the shortest distance are found and then merged into a large cluster until all are merged into one cluster. The whole process is to build a tree structure.

[0085] 1) Sample hierarchical clustering tree

[0086] Hierarchical cluster analysis is performed on each group of samples to form a cluster tree that shows the similarity between samples. The results are as follows: Figure 7 , Figure 8 As shown in Figure 2, samples that are clustered in the same cluster have a higher similarity between them. Figure 7 , Figure 8 Each column represents a sample. Samples with similarity are clustered in the same cluster.

[0087] 2) Hierarchical clustering heat map of significantly different metabolites

[0088] The results of hierarchical cluster analysis of significantly different metabolites (VIP>1, p value<0.05) are as follows Fig. 9 , Fig.10 Metabolites in the same cluster have similar expression patterns and may have similar functions or participate in the same metabolic process or cellular pathway. Fig. 9 , Fig.10 Each row represents a differential metabolite (i.e., the ordinate is the metabolite with significant differential expression), and each column represents a group of samples (i.e., the abscissa is the sample information). Red represents significant upregulation, blue represents significant downregulation, and the color depth indicates the degree of upregulation and downregulation. Metabolites with similar expression patterns are clustered in the same cluster on the left.

[0089] 4.2 Correlation analysis

[0090] Correlation analysis can help measure the metabolic proximities between significantly different metabolites (VIP>1, p value<0.05), which is helpful to further understand the mutual regulatory relationship between metabolites during changes in biological states. Metabolites with expression correlation may participate in a biological process together, that is, functional correlation; in addition, different metabolites have synergistic or mutually exclusive relationships. For example, if the change trend of a certain type of metabolites is the same, it is a positive correlation; if the change trend is opposite to that of a certain type of metabolites, it is a negative correlation. Positively correlated metabolites may also indicate that they are derived from the same synthetic pathway, and negative correlations indicate that they may be decomposed for the synthesis of other metabolites, that is, a synthetic conversion relationship.

[0091] In order to more intuitively reveal the co-regulatory relationship between various metabolites, the correlation matrix was converted into a chord diagram, such as Fig.11 , Fig.12 As shown. The chord diagrams all show metabolite molecular pairs with correlation coefficients |r|>0.8 and p<0.05. This standard can be adjusted according to actual conditions. The chord diagram can better show the correlation between various metabolites, and the network diagram can better show the correlation between various metabolites. Each has its own advantages. Fig.11 , Fig.12 The starting point of the link in the inner circle represents each significantly different metabolite, and the arc on the outer circle represents the category to which the significantly different metabolite belongs. The colored lines represent the correlation within each metabolite, and the lines are the same color as the subclasses. The dark gray lines represent the correlation between metabolites of different categories.

[0092] 5. KEGG analysis of differential metabolites: The screened significant differential metabolites (metabolites must simultaneously meet OPLS-DA VIP>1 and p value<0.05, and have qualitative names) were subjected to subsequent univariate, multivariate and KEGG analysis. Univariate and multivariate analyses showed that these metabolites are mainly involved in biological processes including glycolysis, pentose phosphate pathway, tricarboxylic acid cycle pathway and lipid metabolism, which is consistent with the results of KEGG pathway enrichment bubble analysis showing that metabolites are mainly involved in ABC transporters and tumor central carbon metabolism. In order to facilitate the observation of the expression of each differential metabolite annotated in the KEGG metabolic pathway, the present invention selects a KEGG metabolic pathway with a number of differential metabolites greater than 5, and displays the differential metabolites in the KEGG metabolic pathway in the form of a heat map, such as Fig.13 shown. Fig.13 Each row represents a differential metabolite (i.e., the ordinate is the metabolite with significant differential expression), and each column represents a group of samples (i.e., the abscissa is the sample information). Red represents significant upregulation, blue represents significant downregulation, and the color depth indicates the degree of upregulation and downregulation. Metabolites with similar expression patterns are clustered in the same cluster on the left.

[0093] 6. KEGG pathway enrichment map

[0094] KEGG pathway enrichment analysis uses KEGG pathways as units, and the metabolic pathways involved in the species or closely related species as the background. Fisher's Exact Test is used to analyze and calculate the significance level of metabolite enrichment in each pathway, thereby determining the metabolic and signal transduction pathways that are significantly affected. The smaller the P value, the more significant the difference in the metabolic pathway. The results of metabolic pathway enrichment analysis are displayed in the form of bubble charts and bar charts. The results are shown in Figure 2. Fig.14 and Fig.15 shown. Fig.14 Each bubble in the bubble diagram represents a metabolic pathway (the top 20 with the highest significance are selected based on the p value). The horizontal axis of the bubble and the bubble size indicate the size of the influencing factor of the pathway in the topological analysis. The larger the size, the greater the influencing factor. The vertical axis of the bubble and the bubble color indicate the p-value of the enrichment analysis (negative common logarithm, i.e. -log10 p-value). The darker the color, the smaller the p-value and the more significant the enrichment. The rich factor represents the proportion of the number of differential metabolites in the pathway to the number of metabolites annotated in the pathway. Fig.15The vertical axis in the middle bar graph represents each KEGG metabolic pathway, and the horizontal axis represents the number of differentially expressed metabolites contained in each KEGG metabolic pathway. The color represents the p-value of the enrichment analysis. The darker the color, the smaller the p-value and the more significant the enrichment. The number on the column represents the rich factor, which represents the ratio of the number of differential metabolites in the pathway to the number of metabolites annotated in the pathway.

[0095] Differential Abundance Score is a pathway-based metabolic change analysis method that can capture the average and overall changes of all metabolites in a pathway. The differential abundance scores of all significantly enriched metabolic pathways are as follows: Fig.16 and Fig.17 All significantly enriched metabolic pathways were classified according to their upper level Pathway_Hierarchy, and then re-mapped and displayed. The results are shown in Fig.17 shown. Fig.17 The Y-axis represents the name of the differential pathway, and the X-axis represents the differential abundance score (DA score). The DA score is the overall total change of all metabolites in the metabolic pathway. A score of 1 indicates that the expression trend of all identified metabolites in the pathway is upregulated, and -1 indicates that the expression trend of all identified metabolites in the pathway is downregulated. The length of the line segment represents the absolute value of the DA score, and the size of the dots at the endpoints of the line segment represents the number of metabolites in the pathway. The larger the dots, the more metabolites there are. The depth of the color of the line segment and the dots is proportional to the DA score value. The darker the red, the more the overall expression of the pathway tends to be upregulated, and the darker the blue, the more the overall expression of the pathway tends to be downregulated.

[0096] 7. Quality Control

[0097] 7.1 Quality control (QC) sample correlation

[0098] Pearson correlation analysis was performed on the QC samples (i.e., the clinical samples collected in 1), as shown in Fig.18 and Fig.19 As shown. Generally, a correlation coefficient greater than 0.9 indicates good correlation. The experimental results show that the correlation coefficients between QC samples are all above 0.9, indicating good experimental repeatability. Fig.18 , Fig.19 The horizontal and vertical axes represent each QC sample. The points in each small grid represent the ion peaks (metabolites) extracted from the QC sample, and the horizontal and vertical axes represent the logarithmic values ​​of the ion peak signal intensity values.

[0099] 7.2 Hotelling's T2 Test for Overall Sample

[0100] Hotelling's T2 test tests samples through multivariate modeling and defines 95% or 99% confidence intervals, which can be used to diagnose outlier samples. The results of Hotelling's T2 test are shown in Fig. 20 , Fig.21 The experimental results show that the QC samples are all within the 99% confidence interval, indicating that the repeatability of the experiment is good. Fig. 20 , Fig.21 The horizontal axis represents all experimental samples and QC samples, the vertical axis reflects the confidence interval, and the red line defines the 99% confidence interval range.

[0101] 7.3 Multivariate Control Charts for QC Samples

[0102] The Multivariate Control Chart (MCC) is a multivariate statistical model based on the ion peaks detected by the QC samples. It is a quality management tool used to monitor and determine whether the instrument status is stable. Each point in the multivariate control chart represents a QC sample, and the X-axis is the order in which all QC samples are put on the machine. Due to fluctuations in the instrument status, the points in the chart fluctuate up and down. Generally, the normal range is within plus or minus 3 standard deviations. The multivariate control chart of the QC sample of the present invention is shown in Fig. 22 , Fig.23 The experimental results show that the fluctuations of QC samples are within the range of plus or minus 3 standard deviations, reflecting that the fluctuations of the instrument are within the normal range and the data can be used for subsequent analysis. Fig. 22 , Fig.23 The horizontal axis represents each QC sample, the vertical axis reflects the standard deviation, and the yellow and red lines define the range of plus or minus 2 and 3 standard deviations, respectively.

[0103] 7.4 Relative Standard Deviation (RSD) of QC Samples

[0104] The smaller the relative standard deviation (RSD) of the ion peak abundance of the QC sample, the better the stability of the instrument, which is an important indicator of data quality. The number of peaks with RSD ≤ 30% in the QC sample of this application accounts for more than 80% of the total number of peaks in the QC sample, see Fig.24 , Fig.25 , indicating that the instrument analysis system has good stability and the data can be used for subsequent analysis.

[0105] 8. Proteomic analysis: Proteomic detection of colon adenocarcinoma and colon adenomatous polyps tissues, according to the expression fold change (FC)> 2.0 times (or down-regulation less than 0.5 times) and P value < 0.05 as the standard, 1563 significantly differentially expressed proteins were identified, of which 911 were up-regulated and 652 were down-regulated (such as Fig.26 AB). The differentially expressed proteins were mainly related to the functions of mitochondria and organelle membranes, and were involved in biological processes such as tumor metabolism, regulation of adipocyte lipolysis, glycosaminoglycan biosynthesis-heparin sulfate / heparin, niacin and nicotinamide metabolism (e.g. Fig.26 C). Fig.26 Comparisons in A: differential comparison groups; Significantly changing in abundance: differentially expressed proteins that meet the screening multiples and p value; Consistent presence / absence expression profile: differentially expressed proteins that are not null values ​​twice or more in one group of samples, and all data in another group are null values. Up: up-regulated differentially expressed proteins; Down: down-regulated expressed proteins; Fig.26 In B, the horizontal axis is the difference multiple (logarithmic transformation with base 2), and the vertical axis is the significant P-value of the difference (logarithmic transformation with base 10). The red points in the figure are up-regulated significantly differentially expressed proteins, the blue points are down-regulated significantly differentially expressed proteins, and the gray points are proteins with no difference.

[0106] 9. Cluster analysis of differentially expressed proteins

[0107] In order to analyze the expression patterns of samples between and within groups, test the rationality of the grouping of the present invention, and explain whether the changes in the expression of differential proteins can represent the significant effects of biological treatment on the samples, a hierarchical clustering algorithm was used to group and classify the differentially expressed proteins of the comparison groups and display them in the form of a heatmap. Based on the similarity basis, in the clustering grouping results, the data patterns within the group are generally more similar, while the data patterns between the groups are less similar, so the groups can be effectively distinguished. Fig. 27 , with the screening criteria of fold change > 2.0 times and P value < 0.05 (T-test or other), the significantly differentially expressed proteins obtained can effectively separate the comparison groups, indicating that differentially expressed protein screening can represent the impact of biological treatment on the sample. Fig. 27 The results of mid-level clustering are presented in a tree-like heat map, where each column represents a group of samples (the horizontal axis is sample information), and each row represents a protein (i.e., the vertical axis is the protein with significant differential expression). The expression levels of significantly differentially expressed proteins in different samples are standardized using the Z-Score method and displayed in the heat map in different colors, where red represents significantly up-regulated proteins, blue represents significantly down-regulated proteins, and the gray part represents no protein quantitative information.

[0108] 10. Proteomics quality control

[0109] This application uses a Q Exactive mass spectrometer with high mass accuracy and high resolution, which can maintain good mass deviation during data acquisition and ultimately obtain high-quality MS1 and MS2 spectra. Fig.28 As shown in the figure, the mass deviation of all identified peptides is mainly distributed within 10ppm, indicating that the identification results are accurate and reliable. Then, the MS spectrum data was analyzed with the strict analysis tool Andromeda to obtain the score of each MS2 spectrum. Fig.29 The results show that the Andromeda score of MS2 is relatively ideal, with more than 74.8% of the peptides scoring more than 60 points, and the median peptide score is 88.18 points. Each set of label free data uses peptide FDR≤0.01 and protein FDR≤0.01 as screening criteria in qualitative analysis, and the distribution of excellent peptide scores further indicates that the MS experimental data quality is high. Fig.28 The horizontal axis is the mass deviation between the theoretical mass-to-charge ratio of the peptide ion and the mass-to-charge ratio measured by mass spectrometry experiment. The unit is ppm, which is one part per million, and is a relative unit. The vertical axis is the Andromeda peptide score. Fig.29 The middle horizontal axis is the peptide score; the main vertical axis The Number of Peptides corresponds to the bar graph in the figure, indicating the number of peptides with corresponding ion scores identified; the secondary vertical axis corresponds to the cumulative curve in the figure, indicating the cumulative percentage of peptides not higher than the corresponding ion score.

[0110] 11. Dual-omics joint analysis: Differential proteins and differential metabolites are included in the common pathways, KEGG pathway visualization is implemented, and the expression correlation between proteins and metabolites is globally observed at the two-omics level. By simultaneously projecting the differential proteins obtained by proteomics and the differential metabolites obtained by metabolomics to the KEGG pathway, the significance level of the enrichment of proteins and metabolites in each pathway is calculated, and 30 common pathways (such as Fig.30 The KEGG pathways mainly include retrograde endocannabinoid signaling, 5-hydroxytryptaminergic synapses, glycerophospholipid metabolism, purine metabolism, aminoacyl-tRNA biosynthesis, choline metabolism in cancer, etc. (e.g. Fig.30 Further KEGG pathway enrichment comparison revealed that the cholesterol metabolism biological pathway was extremely active in COAP and COAD, and this pathway may be closely related to the canceration of adenomatous polyps (such as Fig.30 C).

[0111] 12. Screening of differentially expressed genes in TCGA-COAD database:

[0112] Download the Counts matrix data of TCGA-COAD, 440 cases of TCGA-COAD with prognostic information, and 41 cases of normal colon tissue GTEx. Based on the grouping information of the training set, the R software package was used to identify the differential genes between the training set TCGA-COAD and GTEx (limma R package), with cutoff: abs(logFC)>1.5&P.Value<0.01, and 4356 differentially expressed genes were obtained, of which 2765 were up-regulated and 1591 were down-regulated. The distribution of TCGA differentially expressed genes is shown in the figure below: Fig.31 shown. Fig.31 The horizontal axis represents the multiple of difference, taking the logarithm of 2, and the vertical axis represents -log10 (pvalue). Each point in the figure represents a gene, and the green and yellow points represent significantly differentially expressed genes. The red points indicate that the gene expression is upregulated, the blue points indicate that the gene expression is downregulated, and the gray points indicate that there is no significant difference between these genes.

[0113] 13. Joint analysis of proteomics differentially expressed proteins with TCGA database: 1563 significantly differentially expressed proteins were converted to UniProt protein sequence database, and 1559 genes were obtained (4 proteins have no corresponding gene names yet). The differentially expressed genes (ZSXM coherent) of patients with colon cancer combined with multiple adenomatous polyps of the colon were screened and jointly analyzed with the differentially expressed genes in TCGA-COAD database: 530 co-expressed genes were screened, such as Fig.32 shown.

[0114] 14. Differential gene enrichment analysis

[0115] The clusterProfiler R package was used to enrich the 530 differentially co-expressed genes (GO+KEGG). KEGG analysis showed that these differentially expressed genes were involved in chemical carcinogenesis, arginine and proline metabolism, glutathione metabolism, platinum drug resistance, protein digestion and absorption, cell adhesion molecules, non-alcoholic fatty liver disease, ECM-receptor interaction, tight junction and ribosome biogenesis; GO analysis showed that BP components: mitochondrial translation, mitochondrial gene expression, cRNA processing, ribosome biogenesis; CC components: mitochondrial inner membrane, pre-ribosome, mitochondrial protein complex, extracellular matrix, mitochondrial matrix; MF components: integrin binding structural components that bind to extracellular matrix structural components, such as Fig.33 shown.

[0116] 15. Univariate COX regression analysis

[0117] The COX regression model uses survival outcomes and survival time as dependent variables, can simultaneously analyze the impact of many factors on survival, can analyze data with truncated survival time, and does not require the estimation of the survival distribution type of the data. Due to the above excellent properties, the model has been widely used in medical follow-up studies since its inception, and is the most widely used multivariate analysis method in survival analysis to date. The present invention uses the TCGA dataset as a training set, and uses the 530 key genes obtained to perform univariate Cox proportional hazard regression analysis, setting the univariate P.value to 0.05. There are 4 genes in total, namely: FABP4, CSRP2, LRCH4 and TIMP1. The results are shown in Tables 1 and Fig.34 .

[0118] Table 1 Univariate analysis of differentially expressed genes

[0119]

[0120] 16. Prognostic model construction

[0121] The step function was used to perform lasso model construction on the four prognostic genes (FABP4, CSRP2, LRCH4 and TIMP1) in the above results to construct the prognostic score RiskScore, which is the prognostic gene for patients with colon cancer and multiple adenomatous polyps of the colon. The multivariate coefficient (coef) of each prognostic gene was calculated to construct the survival risk model. The prognostic score = 0.270598577*FABP4+0.80024097*CSRP2+2.973438047*LRCH4+1.14225671*TIMP1, as shown in Figure 1. Fig.35 shown.

[0122] 17. Survival risk model evaluation

[0123] In order to evaluate the prognostic value of the risk model, the expression levels of four prognostic genes were obtained in the TCGA dataset, and the risk value of each patient was calculated. The 440 colon cancer patients in the training set were divided into high-risk and low-risk groups, of which there were 118 samples in the high-risk group and 321 samples in the low-risk group. Red represents the high-risk group and blue represents the low-risk group. The results of survival analysis of the high-risk and low-risk groups are shown in Figure 2. Fig.36 As shown in the figure, it can be seen that there is a significant survival difference between the high-risk and low-risk groups in the TCGA data set (p<0.05). Based on the high-risk and low-risk group information, the R software was used to draw the expression heat map of the prognostic genes in the high-risk and low-risk groups. The results are shown in the figure. Fig.37 As shown. Fig.37 In the figure, the horizontal axis is sample information, the vertical axis is differentially expressed genes, red indicates high relative expression, and blue indicates low relative expression.

[0124] 18. Survival risk model validation and evaluation

[0125] RiskScore was calculated using the multi-factor COX model, false positives and true positives were calculated using the survival ROC package, and the results were used to draw the ROC curve and calculate the AUC. Fig.38 As shown in the figure, the entire graph is divided into two parts. The area under the curve is called AUC (Area Under Curve), which is used to indicate prediction accuracy. The higher the AUC value, that is, the larger the area under the curve (and the smoother the curve), the higher the prediction accuracy. The closer the curve is to the upper left corner (the smaller the X, the larger the Y), the higher the prediction accuracy. The AUC values ​​of 1 year, 2 years, 3 years, 4 years and 5 years all reach above 0.6, indicating that the constructed risk regression model can be effectively used as a prognostic model, as shown in the figure. Fig.38 shown.

[0126] In order to verify the applicability of the risk model, the present invention uses the dataset GSE39582 as an external validation set for model validation, obtains the expression levels of 4 prognostic genes in the GSE28722 dataset, and divides the 556 CRC patients in the dataset GSE39582 into high-risk and low-risk groups, of which there are 272 samples in the high-risk group and 284 samples in the low-risk group. Red represents the high-risk group and blue represents the low-risk group. The ROC curve is drawn, and the AUC values ​​of 1 year, 2 years, 3 years, 4 years and 5 years are all above 0.6, indicating that the constructed risk regression model can be effectively used as a prognostic model, such as Fig.39 The results of survival analysis of high-risk and low-risk groups using TCGA and GSE39582 databases are shown in Fig.40 and Fig.41 As shown in the figure, it can be seen that there is a significant survival difference between the high-risk and low-risk groups in the external validation set (p<0.05). Fig.40 and 41 The vertical axis represents the survival rate, the horizontal axis represents the overall survival time (OS), the red curve represents the high-risk group, and the blue curve represents the low-risk group. The difference between the high-risk and low-risk groups is less than 0.05, which is a significant difference.

[0127] 19. Expression of prognostic genes in colon and prognostic analysis

[0128] Based on the obtained 4 prognostic genes: FABP4, CSRP2, LRCH4 and TIMP1, the expression matrix in the TCGA training set and the expression matrix in the external validation set GSE39582 were extracted respectively. Combined with the grouping information of the disease group and the normal group samples, the R software package "ggplot2" was used to draw the expression bar graphs of the prognostic genes in the two data sets, and Wilcoxon was used for verification. It was found that the 4 prognostic genes had significant differences between the disease group and the normal control group in the training set, and the expression levels in the disease group were higher than those in the normal control group, which was consistent with the results of the difference analysis. The 4 prognostic genes were up-regulated genes, such as Fig.42 shown.

[0129] 20. Single gene enrichment of prognostic genes

[0130] In the TCGA training set, the present invention calculates the Pearson correlation coefficient between each prognostic gene and all genes in the data set, and then sorts the correlation coefficients from large to small. After each key gene obtains the corresponding gene ranking list, the R package "clusterProfiler" is used to perform KEGG GSEA enrichment analysis, and the Benjamini&Hochberg method is used for multiple test correction to obtain the corrected p value, i.e., p.adjust. p.adjust<0.05 is considered a significant enrichment result. According to the significance ranking, KEGG TOP6 Descriptions are selected for visual display, involving colon cancer, cholesterol metabolism, ECM receptor interaction, focal adhesion, Th1 and Th2 cell differentiation, and Ras signaling pathways, such as Fig.43 shown.

[0131] 21. Patient’s immune status

[0132] In order to explore the immune characteristics of the prognostic model, ESTIMATE was used to calculate the matrix score, tumor purity, immune score and ESTIMATE score of TCGA-COAD samples. The risk score calculated by the prognostic model was correlated with the above scores. The results showed that the risk score was positively correlated with the matrix score, immune score and ESTIMATE score. The immune, matrix and ESTIMATE scores of high-risk patients were higher than those of low-risk patients (P<0.05, as shown in Table 2). Fig.44 shown).

[0133] 22. Immune status of risk model: The ESTIMATE (Estimation of stromal and immune cells in malignant tumor tissues using expressions) algorithm was used to estimate immune cell abundance (immune score), stromal cell infiltration level (stromal score), and tumor purity (ESTIMATE score). A comprehensive analysis of immune infiltration in the colon cancer microenvironment was performed using the Tumor Immunity Evaluation Resource 2.0 (TIMER 2.0). The MCPcounter algorithm was used to estimate the relative proportions of 10 immune cells in colon cancer. The above algorithm shows the abundance of infiltrating immune cell populations under high and low risks of different risk models, such as Fig.45 shown.

[0134] 23. Analysis of clinical characteristics and immune checkpoints in the prognostic model

[0135] In order to analyze the correlation between risk score expression and clinical characteristics, the differences in patient risk scores were compared according to the clinical characteristics of different groups. The results showed that the risk scores of N stage, T stage, M stage and Stage status were statistically significant. The heat map of risk model and clinical characteristics ( Fig.46 ). Immune checkpoints are a class of immunosuppressive molecules that are expressed on immune cells and can regulate the degree of immune activation. They play an important role in preventing the occurrence of autoimmune effects. In layman's terms, they are small protein molecules produced by immune cells that regulate autoimmune functions. Immune checkpoint drugs have been shown to have anti-tumor effects that reverse tumor immunosuppression, so the applicant analyzed the correlation between the risk score and the expression levels of different immune checkpoint proteins (MICB, ITGB2, BTN3A2, ICOSLG, CX3CL1, PRF1, CTLA4, etc.). The results showed that the risk score was significantly correlated with these checkpoints (such as Fig.46 shown). Fig.46 *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001

[0136] 24. Prognostic model and tumor mutation burden analysis

[0137] Based on the prognostic model, the present invention performs mutation burden analysis on the high-risk group and low-risk group of the TCGA-COAD cohort. From the waterfall plot ( Fig.47 and Fig.48), it was found that the top 30 highly mutated genes in the high-risk group were APC, TP53, TTN, KRAS, MUC16, SYNE1, PIK3CA, RYR2, PCLO, USH2A, FAT4, BRAF, KMT2D, OBSCN, RYR1, ZFHX4, DNAH11, MUC5B, NEB, RYR3, ABCA13, CCDC168, CSMD3, DNAH5, SDK1, VPS13B, XIRP2, DNAH10, KIAA1109, and PCDH15; the low-risk group The top 30 highly mutated genes in the two groups were APC, TP53, TTN, KRAS, PIK3CA, MUC16, SYNE1, FAT4, OBSCN, RYR2, ZFHX4, DNAH5, FAT3, LRP1B, CSMD3, ABCA13, PCLO, CSMD1, DNAH11, LRP2, SOX9, HMCN1, ADGRV1, DMD, FBXW7, FLG, USH2A, ATM, CCDC168 and NEB. The mutation ratios of each highly mutated gene were not the same between the two groups.

[0138] 25. Tumor drug sensitivity analysis

[0139] The Genomics of Drug Sensitivity in Cancer (GDSC) is the largest public resource on cancer cell drug sensitivity and drug response molecular markers. The present invention first downloads drug sensitivity data from the GDSC database, uses the "oncoPredict" R program to calculate the IC50 of each colon cancer patient between high-risk and low-risk groups, and obtains a total of 51 drugs after filtering the data through "clinical trial validation and FDA standard certification". Fig.49 The IC50 box plot of some treatment-sensitive drugs was displayed. The results showed that the IC50 of the high-risk group was significantly lower than that of the low-risk group.

[0140] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0141] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above invention concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A method for constructing a prognostic model for patients with colon adenocarcinoma combined with multiple adenomatous polyps, characterized in that: The following steps are involved: S1. Collect clinical samples of colon adenocarcinoma and clinical samples of multiple adenomatous polyps, perform metabolomics detection on them respectively, obtain metabolites, perform differential analysis on the metabolites, and screen out significantly differential metabolites according to a set first threshold; S2, performing proteomic analysis on the clinical samples of colon adenocarcinoma and multiple adenomatous polyps, and screening out significantly differentially expressed proteins according to a set second threshold; S3. After jointly analyzing the significantly differential metabolites and the significantly differential proteins, differential genes are screened out based on the TCGA-COAD database and the GTEx database, the significantly differential proteins are converted into differential protein genes, and the differential protein genes are jointly analyzed with the differential genes to obtain co-expressed genes; S4. Using the co-expressed genes to perform univariate Cox proportional hazard regression analysis to identify prognostic genes; S5. Constructing a prognostic model according to the calculated multi-factor coefficients of the prognostic genes.

2. The method for constructing a prognostic model for patients with colon adenocarcinoma combined with multiple adenomatous polyps according to claim 1, characterized in that: In step S1, multiple significantly differential metabolites were screened out according to expression fold>1.5 or expression fold<0.67, OPLS-DA VIP>1 and statistical significance P value<0.05; In step S2, multiple significantly differentially expressed proteins were screened out according to expression fold>2.0 or expression fold<0.5, and statistically significant P value<0.

05.

3. The method for constructing a prognostic model for patients with colon adenocarcinoma combined with multiple adenomatous polyps according to claim 1, characterized in that: In step S1, the metabolites are differentially analyzed by multidimensional analysis, bioinformatics analysis and metabolomics experimental data quality control, wherein the multidimensional analysis includes principal component analysis of the metabolites and detection of the metabolites by a partial least squares discriminant analysis model, the bioinformatics analysis includes cluster analysis, correlation analysis and pathway analysis, and the metabolomics experimental data quality control includes quality control sample correlation, overall sample inspection, multivariate control and relative standard deviation control.

4. The method for constructing a prognostic model for patients with colon adenocarcinoma combined with multiple adenomatous polyps according to claim 1, characterized in that: In step S3, the significantly different metabolites and the significantly different proteins are jointly analyzed, including: projecting the significantly different metabolites and the significantly different proteins to the KEGG pathway at the same time, and calculating the enrichment significance level of the significantly different metabolites and the significantly different proteins in each pathway. The KEGG pathway includes: retrograde endocannabinoid signaling, serotonergic synapses, glycerophospholipid metabolism, purine metabolism, aminoacyl-tRNA biosynthesis and choline metabolism in cancer.

5. The method for constructing a prognostic model for patients with colon adenocarcinoma and multiple adenomatous polyps according to claim 1, characterized in that: In step S3, it also includes: converting the significantly differentially expressed proteins in the UniProt protein sequence database to obtain differentially expressed protein genes, jointly analyzing the differentially expressed protein genes with differentially expressed genes in the TCGA-COAD database, and screening to obtain multiple co-expressed genes.

6. The method for constructing a prognostic model for patients with colon adenocarcinoma and multiple adenomatous polyps according to claim 1, characterized in that: The membrane protein R language package was used to perform enrichment analysis on the co-expressed genes, including GO analysis and KEGG analysis.

7. The method for constructing a prognostic model for patients with colon adenocarcinoma and multiple adenomatous polyps according to claim 1, characterized in that: In step S4, the TCGA dataset was used as a training set, and the co-expressed genes obtained were used to perform univariate Cox proportional hazard regression analysis, and the univariate P value was set to 0.05 to identify prognostic genes.

8. The method for constructing a prognostic model for patients with colon adenocarcinoma and multiple adenomatous polyps according to claim 1, characterized in that: The prognostic genes include: FABP4, CSRP2, LRCH4 and TIMP1.

9. The method for constructing a prognostic model for patients with colon adenocarcinoma and multiple adenomatous polyps according to claim 1, characterized in that: In step S5, it also includes: using the step function to construct a prognostic score for the prognostic gene using the lasso model, calculating the multi-factor coefficient of each prognostic gene for constructing the prognostic model, and verifying and evaluating the prognostic model.

10. A prognostic model for patients with colon adenocarcinoma and multiple adenomatous polyps, characterized in that: The method is constructed according to any one of claims 1 to 9.