A method and system for processing hepatocellular carcinoma data

Abstract

This invention provides a method and system for processing hepatocellular carcinoma data. The method utilizes the CIBERSORT tool combined with weighted gene co-expression network analysis to screen a gene set positively correlated with M2 macrophage infiltration. Through differential expression analysis and prognostic correlation analysis, target genes with prognostic value are identified from the gene set. A dataset containing the target genes is acquired, and a comprehensive machine learning algorithm is used to generate several corresponding prognostic prediction models for each dataset. The C-index of the dataset in the corresponding prognostic prediction model is calculated, and target prognostic prediction models are selected based on the C-index and the complexity of the prognostic prediction model. Compared to the traditional method of selecting an algorithm for modeling at the beginning of the study, the final selected target prognostic prediction model reflects more objective and realistic prediction results. Furthermore, the prediction accuracy of this target prognostic prediction model is improved.

Description

Technical Field

[0001] This invention belongs to the field of biological data analysis technology, and specifically relates to a method and system for processing hepatocellular carcinoma data. Background Technology

[0002] Hepatocellular carcinoma (HCC) is one of the most common malignant tumors. The current epidemiological status of liver cancer is concerning, necessitating new breakthroughs in all aspects of HCC prevention, diagnosis, treatment, and prognostic assessment.

[0003] It should be noted that most prognostic prediction models built on machine learning theory use predetermined algorithms, which makes the prediction results highly subjective and often fails to achieve the model's maximum predictive performance. Summary of the Invention

[0004] Based on this, the present invention provides a method and system for processing hepatocellular carcinoma data, aiming to establish a prognostic model and improve prediction accuracy.

[0005] A first aspect of this invention provides a method for processing hepatocellular carcinoma data, the method comprising:

[0006] Using the CIBERSORT tool combined with weighted gene co-expression network analysis, we screened gene sets that were positively correlated with M2 macrophage infiltration.

[0007] Target genes with prognostic value were identified from the gene set through differential expression analysis and prognostic correlation analysis.

[0008] A dataset containing the target gene is obtained, and a comprehensive machine learning algorithm is used to generate several corresponding prognostic prediction models in each dataset.

[0009] Calculate the C-index of each prognostic prediction model under different datasets, calculate the average C-index, and determine the prognostic prediction model corresponding to the maximum average C-index.

[0010] Determine whether the prognostic prediction model corresponding to the maximum average value of the C-index is unique;

[0011] If so, the prognostic prediction model corresponding to the maximum average value of C-index is determined as the target prognostic prediction model;

[0012] If not, then based on the C-index of each prognostic prediction model under different datasets, calculate the complexity of the prognostic prediction model corresponding to the maximum average C-index, and determine the prognostic prediction model corresponding to the minimum complexity as the target prognostic prediction model.

[0013] Furthermore, the datasets are RNA-seq datasets for hepatocellular carcinoma users, including the TCGA dataset, ICGC dataset, GSE14520 dataset, and GSE76427 dataset.

[0014] Furthermore, the step of using the CIBERSORT tool combined with weighted gene co-expression network analysis to screen for gene sets positively correlated with M2 macrophage infiltration includes:

[0015] The overall RNA sequencing data from the TCGA dataset was input into the CIBERSORT tool to generate an immune cell infiltration expression matrix for each hepatocellular carcinoma sample in the corresponding dataset.

[0016] We used weighted gene co-expression network analysis to analyze the immune cell infiltration expression matrix of each hepatocellular carcinoma sample in the dataset;

[0017] The neighborhood relationships were converted into topological overlap matrices, and the average values ​​of different topological overlap matrix measures were used to cluster them into chain hierarchical structures. Gene sets that are positively correlated with M2 macrophage infiltration were then identified.

[0018] Furthermore, in the step of analyzing the immune cell infiltration expression matrix of each hepatocellular carcinoma sample in the dataset using weighted gene co-expression network analysis, the independence β value is set to 0.9, and hepatocellular carcinoma samples with an average expression level >0.5 are selected for analysis.

[0019] Furthermore, the step of identifying target genes with prognostic value from the gene set through differential expression analysis and prognostic correlation analysis includes: using transcriptome expression differential analysis, with |LogFC|≥1.5 as a threshold, identifying the first differentially expressed gene in the TCGA dataset between hepatocellular carcinoma tissue and normal liver tissue, where |LogFC| is the absolute value of the logarithmic fold change;

[0020] Identify the second differentially expressed gene between hepatocellular carcinoma tissue and normal liver tissue by finding the same members of the first differentially expressed gene and M2 macrophages.

[0021] Cox regression analysis was used to compare the correlation between the expression of the second differentially expressed gene and the overall survival time of hepatocellular carcinoma patients to identify the target gene.

[0022] Furthermore, the comprehensive machine learning algorithm includes 8 single algorithms and 22 joint algorithms, including SurvivalSVM, SuperPC, Enet, StepCox, Ridge Regression, plsRcox, RSF, and GBM.

[0023] Furthermore, the step of calculating the complexity of the prognostic prediction model corresponding to the maximum average C-index of each prognostic prediction model under different datasets includes:

[0024] Determine the likelihood function of the prognostic prediction model corresponding to the maximum average value of the C-index. The likelihood function includes the parameter vector of the prognostic prediction model.

[0025] Take the logarithm of the likelihood function to obtain the log-likelihood function. Then, take the derivative with respect to the parameter vector and set the derivative to zero to find the parameter estimate that maximizes the log-likelihood function, thus obtaining the maximum likelihood estimate.

[0026] The number of parameters in the prognostic prediction model corresponding to the maximum average value of the C-index is determined, and the complexity is calculated based on the number of parameters and the maximum likelihood estimate.

[0027] The C-index of each prognostic prediction model under different datasets is ranked from largest to smallest. Based on the ranking results and the importance of the datasets, the adjustment factor corresponding to the C-index of each prognostic prediction model under different datasets is determined.

[0028] The target complexity is calculated based on the adjustment factor and the complexity.

[0029] A second aspect of this invention provides a hepatocellular carcinoma data processing system for implementing the hepatocellular carcinoma data processing method provided in the first aspect, the system comprising:

[0030] The screening module is used to use the CIBERSORT tool combined with weighted gene co-expression network analysis to screen gene sets that are positively correlated with M2 macrophage infiltration;

[0031] The first determination module is used to determine target genes with prognostic value from the gene set through differential expression analysis and prognostic correlation analysis.

[0032] The generation module is used to acquire a dataset containing the target gene and to generate several corresponding prognostic prediction models in each dataset using a comprehensive machine learning algorithm.

[0033] The calculation module is used to calculate the C-index of each prognostic prediction model under different datasets, calculate the average value of the C-index, and determine the prognostic prediction model corresponding to the maximum average value of the C-index.

[0034] The judgment module is used to determine whether the prognostic prediction model corresponding to the maximum average value of C-index is unique.

[0035] The second determining module is used to determine the prognostic prediction model corresponding to the maximum average value of C-index as the target prognostic prediction model if the prognostic prediction model corresponding to the maximum average value of C-index is unique.

[0036] The third determination module is used to calculate the complexity of the prognostic prediction model corresponding to the maximum average value of the C-index based on the C-index of each prognostic prediction model under different datasets if the prognostic prediction model corresponding to the maximum average value of the C-index is not unique, and determine the prognostic prediction model corresponding to the minimum complexity value as the target prognostic prediction model.

[0037] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the hepatocellular carcinoma data processing method provided in the first aspect.

[0038] A fourth aspect of the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the program to implement the hepatocellular carcinoma data processing method provided in the first aspect.

[0039] This invention provides a method and system for processing hepatocellular carcinoma data. The method utilizes the CIBERSORT tool combined with weighted gene co-expression network analysis to screen a gene set positively correlated with M2 macrophage infiltration. Through differential expression analysis and prognostic correlation analysis, target genes with prognostic value are identified from the gene set. A dataset containing the target genes is acquired, and a comprehensive machine learning algorithm is used to generate several corresponding prognostic prediction models for each dataset. The C-index of the dataset in the corresponding prognostic prediction model is calculated, and target prognostic prediction models are selected based on the C-index and the complexity of the prognostic prediction model. Compared to the traditional method of selecting an algorithm for modeling at the beginning of the study, the final selected target prognostic prediction model reflects more objective and realistic prediction results. Furthermore, the prediction accuracy of this target prognostic prediction model is improved. Attached Figure Description

[0040] Figure 1 This is a flowchart illustrating the implementation of a hepatocellular carcinoma data processing method according to Embodiment 1 of the present invention.

[0041] Figure 2 This is a structural block diagram of a hepatocellular carcinoma data processing system provided in Embodiment 2 of the present invention;

[0042] Figure 3 This is a structural block diagram of an electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0043] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0044] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0045] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0046] Example 1

[0047] Please see Figure 1 , Figure 1 The following is a flowchart illustrating the implementation of a hepatocellular carcinoma data processing method according to Embodiment 1 of the present invention. The hepatocellular carcinoma data processing method specifically includes steps S01 to S07.

[0048] Step S01: Using the CIBERSORT tool combined with weighted gene co-expression network analysis, a gene set positively correlated with M2 macrophage infiltration was screened.

[0049] It should be noted that samples of hepatocellular carcinoma were collected first. The inclusion criteria for the samples were as follows: (1) patients diagnosed with HCC by histopathology; (2) patients with a follow-up period of more than 3 months and complete follow-up information. The exclusion criteria were as follows: (1) HCC samples from duplicate patients; (2) samples from patients with metastatic HCC; (3) samples from patients who received anti-HCC therapy before surgery; (4) samples from patients with missing clinical data; (5) samples from patients with missing RNA sequencing data.

[0050] In this embodiment, in order to establish a target prognostic prediction model, four RNA-seq datasets containing large sample sizes of HCC patients were collected, including the TCGA dataset (n = 329), the ICGC dataset (n = 228), the GSE14520 dataset (n = 218), and the GSE76427 dataset (n = 90), where n represents the number of samples.

[0051] The steps of using the CIBERSORT tool combined with weighted gene co-expression network analysis to screen for gene sets positively correlated with M2 macrophage infiltration include:

[0052] The overall RNA sequencing data from the TCGA dataset was input into the CIBERSORT tool to generate an immune cell infiltration expression matrix for each hepatocellular carcinoma sample in the corresponding dataset.

[0053] We used weighted gene co-expression network analysis to analyze the immune cell infiltration expression matrix of each hepatocellular carcinoma sample in the dataset;

[0054] The neighborhood relationships were converted into topological overlap matrices, and the average values ​​of different topological overlap matrix measures were used to cluster them into chain hierarchical structures. Gene sets that are positively correlated with M2 macrophage infiltration were then identified.

[0055] Specifically, CIBERSORT is a tool that deconvolves the expression matrix of human immune cell subtypes based on the principle of linear support vector regression. It provides gene expression feature sets for 22 immune cell subtypes by default, including plasma cells, naive B cells, CD8+ T cells, M0 macrophages, M2 macrophages, dormant natural killer cells, activated natural killer cells, activated CD4+ T cells, dormant CD4+ T cells, M1 macrophages, memory B cells, activated mast cells, dormant mast cells, neutrophils, naive CD4+ T cells, activated dendritic cells, dormant dendritic cells, Gamma-delta T cells, regulatory T cells, eosinophils, monocytes, and follicular helper T cells.

[0056] Weighted correlation network analysis (WGCNA) is a systematic statistical method used to describe gene associations between different samples. It is often used to identify gene sets with highly covariant expression and to screen potential disease biomarkers and therapeutic targets based on the internal connections of gene sets and the association between gene sets and selected phenotypes.

[0057] To more accurately describe the correlation between gene sets and phenotypes, a β value of 0.9 was set during network construction, and samples with average expression levels > 0.5 were selected for analysis. Neighborhood relationships were then converted into topological overlap matrices (TOMs), and clustered into hierarchical chains based on the average values ​​of different TOM metrics. Genes in the modules that were significantly positively correlated with macrophage M2 were defined as M2-type macrophage-related genes.

[0058] Step S02: Identify target genes with prognostic value from the gene set through differential expression analysis and prognostic correlation analysis.

[0059] Specifically, through transcriptome expression differential analysis, with |LogFC|≥1.5 as the threshold, the first differentially expressed gene in the TCGA dataset between hepatocellular carcinoma tissue and normal liver tissue was identified. Here, |LogFC| is the absolute value of the logarithmic fold change, and FC is the fold change, which is calculated by dividing the average expression level of the gene in the experimental group by the average expression level of the gene in the control group.

[0060] Identify the second differentially expressed gene between hepatocellular carcinoma tissue and normal liver tissue by finding the same members of the first differentially expressed gene and M2 macrophages.

[0061] Cox regression analysis was used to compare the expression of the second differentially expressed gene with the overall survival time of hepatocellular carcinoma patients to identify the target genes, which were VASP, TRMU, TRIM28, TIMM50, TBCB, SUPT5H, SNRPA, SNAI2, SMYD3, REEP4, RECQL4, RANBP1, PTK7, PSMD8, PRMT1, PPAN, PAF1, LAPTM4B, HMGA1, GRWD1, ERCC2, EIF4EBP1, CREB3L1, CPNE1, CD320, BCAT1, AKR1B1, and ACTN1.

[0062] Step S03: Obtain a dataset containing the target gene, and use a comprehensive machine learning algorithm to generate several corresponding prognostic prediction models in each dataset.

[0063] Specifically, the target genes and their corresponding patient survival data from the TCGA, ICGC, GSE14520, and GSE76427 datasets were incorporated into a "comprehensive machine learning algorithm" module. Specifically, this module generated 30 prognostic prediction models for each of the four datasets. For example, the prediction model corresponding to the StepCox[backward]+SuperPC method had the highest average C-index among the four datasets, at 0.64, meaning the target prognostic prediction model could be the prediction model corresponding to the StepCox[backward]+SuperPC method.

[0064] Understandably, comprehensive machine learning algorithms include 8 individual algorithms and 22 joint algorithms, including SurvivalSVM, SuperPC, Enet, StepCox, Ridge Regression, plsRcox, RSF, and GBM. Specifically, these include: StepCox[backward]+SuperPC, StepCox[both]+survivalSVM, StepCox[backward]+survivalSVM, survivalSVM, SuperPC, StepCox[both]+SuperPC, RSF+survivalSVM, StepCox[backward]+Enet[alpha=0.1], Enet[alpha=0.1], StepCox[both]+Ridge, StepCox[backward]+Ridge, and RSF+StepCox[forward]. The algorithms listed above are: RSF+plsRcox, RSF+SuperPC, Ridge, RSF+Ridge, StepCox[backward]+Enet[alpha=0.1], StepCox[both], StepCox[backward], RSF, StepCox[forward], StepCox[backward]+RSF, StepCox[both]+plsRcox, StepCox[backward]+plsRcox, RSF+GBM, GBM, StepCox[both]+RSF, StepCox[both]+GBM, plsRcox, StepCox[backward]+GBM. Among these, RSF and StepCox algorithms can be used for feature selection (i.e., selecting meaningful features from all features that are helpful to the model, avoiding the need to import all features into the model for training). Therefore, each of the above joint algorithms includes at least one of the RSF and StepCox algorithms.

[0065] Based on the selected target prognostic prediction model, a risk score can be assigned to each hepatocellular carcinoma (HCC) user across four datasets. It should be noted that only 12 of the 28 M2 macrophage-related genes were used in the model's risk score construction. The formula for calculating the risk score for each HCC sample is: Risk Score = 0.5671879 × SNRPA expression level + (-0.4282688) × RECQL4 expression level + 0.6410663 × TRIM28 expression level + (-0.3376426) × ERCC2 expression level + 0.3590617 × GRWD1 expression level + 0.4491523 × REEP4 expression level + (-1.2190593) × PRMT1 expression level + (-0.3341503) ×PAF1 expression level + 0.2864963 × LAPTM4B expression level + 0.3805174 × RANBP1 expression level + (-0.3052321) × BCAT1 expression level + 0.4239179 × PTK7 expression level. Next, using the optimal cutoff value from the survminer package in R software as the dividing point, the HCC samples were divided into high-risk and low-risk groups.

[0066] Step S04: Calculate the C-index of each prognostic prediction model under different datasets, calculate the average value of the C-index, and determine the prognostic prediction model corresponding to the maximum average value of the C-index.

[0067] For example, StepCox[backward]+SuperPC calculates a C-index of 0.683 on the TCGA dataset, 0.692 on the ICGC dataset, 0.530 on the GSE14520 dataset, and 0.644 on the GSE76427 dataset; StepCox[both]+survivalSVM calculates a C-index of 0.671 on the TCGA dataset and a C-index of 0.530 on the ICGC dataset. The C-index calculated for the StepCox[backward]+survivalSVM dataset is 0.694 in the TCGA dataset, 0.694 in the ICGC dataset, 0.544 in the GSE14520 dataset, and 0.639 in the GSE76427 dataset.

[0068] It can be observed that the average C-index of the StepCox[backward]+SuperPC joint algorithm, the StepCox[both]+survivalSVM joint algorithm, and the StepCox[backward]+survivalSVM joint algorithm is 0.64, which is the prognostic prediction model corresponding to the maximum average C-index.

[0069] Step S05: Determine whether the prognostic prediction model corresponding to the maximum average value of C-index is unique. If yes, proceed to step S06; otherwise, proceed to step S07.

[0070] In step S06, the prognostic prediction model corresponding to the maximum average value of C-index is determined as the target prognostic prediction model.

[0071] In step S07, based on the C-index of each prognostic prediction model under different datasets, the complexity of the prognostic prediction model corresponding to the maximum average value of C-index is calculated, and the prognostic prediction model corresponding to the minimum complexity value is determined as the target prognostic prediction model.

[0072] Specifically, we determine the likelihood function of the prognostic prediction model corresponding to the maximum average value of the C-index. The likelihood function includes the parameter vector of the prognostic prediction model. Understandably, different models have different forms of likelihood functions. For example, for a linear regression model, assuming the error follows a normal distribution, its likelihood function can be expressed as:

[0073] ;

[0074] Where θ is the parameter vector of the model. y=(y 1 ,y 2 ,...,y n ) X is the observed response variable vector, and X is the independent variable matrix. It is the first prediction based on the model. i One observation value, σ 2 It is the variance of the error, where n is the sample size, and y is the variance of the error. i For the first i Each observation value. For other models, such as logistic regression and survival analysis models, there are also corresponding likelihood function expressions, which can usually be derived based on the probability distribution assumptions of the model.

[0075] Taking the logarithm of the likelihood function yields the log-likelihood function. Then, differentiating the parameter vector and setting the derivative to zero, we find the parameter estimates that maximize the log-likelihood function, thus obtaining the maximum likelihood estimate. For example, the log-likelihood function for the linear regression model described above is:

[0076] ;

[0077] The parameter estimates are obtained by solving the problem. Then, substituting it into the log-likelihood function, we obtain the maximum log-likelihood value. .

[0078] The number of parameters in the prognostic prediction model corresponding to the maximum average C-index is determined. Based on this number of parameters and the maximum likelihood estimate, the complexity is calculated. It should be noted that the number of parameters is denoted as... k， Understandably, the number of parameters is related to the dataset to be used; the larger the dataset, the more parameters are required. For example, in a linear regression model... In, the number of parameters k = p +1, Let p be the p-th independent variable. Let be the regression coefficient corresponding to the p-th independent variable. For more complex models, such as those containing interaction terms, polynomial terms, or hierarchical structures, the number of independent parameters needs to be carefully calculated. Furthermore, taking a linear regression model as an example, according to the formula... The computational complexity is 2, where 2 k It is a penalty term for model complexity. It reflects how well the model fits the data. AIC represents complexity. The lower the complexity, the better the model has achieved a balance between goodness of fit and complexity, and the better the model performance.

[0079] The C-index of each prognostic prediction model under different datasets is ranked from largest to smallest. Based on the ranking results and the importance of the dataset, the adjustment factor corresponding to the C-index of each prognostic prediction model under different datasets is determined. Specifically, the mapping relationship between ranking, importance of dataset and adjustment factor is established in advance. That is, the combination of ranking A and importance B corresponds to adjustment factor C. It should be noted that the importance of dataset can be determined by expert evaluation.

[0080] The target complexity is calculated based on the adjustment factor and the complexity. There are four datasets, which means there are four C-index rankings and four corresponding adjustment factors. The adjustment factors are summed and multiplied by the complexity to obtain the target complexity. Finally, the prognostic prediction model corresponding to the minimum complexity is determined as the target prognostic prediction model.

[0081] In summary, the hepatocellular carcinoma data processing method in the above embodiments of the present invention utilizes the CIBERSORT tool combined with weighted gene co-expression network analysis to screen a gene set positively correlated with M2 macrophage infiltration; through differential expression analysis and prognostic correlation analysis, target genes with prognostic value are identified from the gene set; a dataset containing the target genes is obtained; a comprehensive machine learning algorithm is used to generate several corresponding prognostic prediction models in each dataset; the C-index of the dataset in the corresponding prognostic prediction model is calculated; and target prognostic prediction models are selected based on the C-index and the complexity of the prognostic prediction model. The final selected target prognostic prediction model, compared to the traditional method of selecting an algorithm for modeling at the beginning of the study, reflects more objective and realistic prediction results, and the prediction accuracy of this target prognostic prediction model is improved.

[0082] Example 2

[0083] Please see Figure 2 , Figure 2 This is a structural block diagram of a hepatocellular carcinoma data processing system provided in Embodiment 2 of the present invention. The hepatocellular carcinoma data processing system 200 includes: a screening module 21, a first determination module 22, a generation module 23, a calculation module 24, a judgment module 25, a second determination module 26, and a third determination module 27, wherein:

[0084] Screening module 21 is used to screen gene sets that are positively correlated with M2 macrophage infiltration using the CIBERSORT tool combined with weighted gene co-expression network analysis;

[0085] The first determining module 22 is used to determine target genes with prognostic value from the gene set through differential expression analysis and prognostic correlation analysis.

[0086] The generation module 23 is used to obtain a dataset containing the target gene and to generate several corresponding prognostic prediction models in each dataset using a comprehensive machine learning algorithm. The dataset is the RNA-seq dataset of hepatocellular carcinoma users, including the TCGA dataset, ICGC dataset, GSE14520 dataset and GSE76427 dataset. The comprehensive machine learning algorithm includes 8 single algorithms and 22 joint algorithms, including SurvivalSVM, SuperPC, Enet, StepCox, Ridge regression, plsRcox, RSF and GBM.

[0087] The calculation module 24 is used to calculate the C-index of each prognostic prediction model under different datasets, calculate the average value of the C-index, and determine the prognostic prediction model corresponding to the maximum average value of the C-index.

[0088] Module 25 is used to determine whether the prognostic prediction model corresponding to the maximum average value of C-index is unique;

[0089] The second determining module 26 is used to determine the prognostic prediction model corresponding to the maximum average value of C-index as the target prognostic prediction model if the prognostic prediction model corresponding to the maximum average value of C-index is unique.

[0090] The third determining module 27 is used to calculate the complexity of the prognostic prediction model corresponding to the maximum average value of C-index based on the C-index of each prognostic prediction model under different datasets if the prognostic prediction model corresponding to the maximum average value of C-index is not unique, and to determine the prognostic prediction model corresponding to the minimum complexity value as the target prognostic prediction model.

[0091] Furthermore, in other embodiments of the present invention, the screening module 21 includes:

[0092] The generation unit is used to bring the overall RNA sequencing data in the TCGA dataset into the CIBERSORT tool to generate the immune cell infiltration expression matrix for each hepatocellular carcinoma sample in the corresponding dataset.

[0093] The analysis unit is used to analyze the immune cell infiltration expression matrix of each hepatocellular carcinoma sample in the dataset using weighted gene co-expression network analysis. The independence β value is set to 0.9, and hepatocellular carcinoma samples with an average expression level >0.5 are selected for analysis.

[0094] The transformation unit is used to convert neighborhood relationships into topological overlap matrices, cluster them into chain hierarchical structures based on the average values ​​of different topological overlap matrix measures, and identify gene sets that are positively correlated with M2 macrophage infiltration.

[0095] Furthermore, in other embodiments of the present invention, the first determining module 22 includes:

[0096] The first determining unit is used to identify the first differentially expressed gene between hepatocellular carcinoma tissue and normal liver tissue in the TCGA dataset by means of transcriptome expression differential analysis, with |LogFC|≥1.5 as the threshold, where |LogFC| is the absolute value of the logarithmic fold change;

[0097] The second determining unit is used to find the same members of the first differentially expressed gene and M2 macrophages, and to determine the second differentially expressed gene between hepatocellular carcinoma tissue and normal liver tissue.

[0098] The third determining unit is used to use Cox regression analysis to compare the correlation between the expression of the second differentially expressed gene and the overall survival time of hepatocellular carcinoma users, and to determine the target gene.

[0099] Furthermore, in other embodiments of the present invention, the third determining module 27 includes:

[0100] The first determining unit is used to determine the likelihood function of the prognostic prediction model corresponding to the maximum average value of C-index, and the likelihood function includes the parameter vector of the prognostic prediction model.

[0101] The first calculation unit is used to take the logarithm of the likelihood function to obtain the log-likelihood function, then take the derivative of the parameter vector and set the derivative to zero to find the parameter estimate that maximizes the log-likelihood function, and obtain the maximum likelihood estimate.

[0102] The second calculation unit is used to determine the number of parameters of the prognostic prediction model corresponding to the maximum average value of C-index, and calculate the complexity based on the number of parameters and the maximum likelihood estimate.

[0103] The second determining unit is used to rank the C-index of each prognostic prediction model under different datasets according to the numerical values ​​from large to small, and to determine the adjustment factor corresponding to the C-index of each prognostic prediction model under different datasets based on the ranking results and the importance of the datasets.

[0104] The third calculation unit is used to calculate the target complexity based on the adjustment factor and the complexity.

[0105] Example 3

[0106] In another aspect, the present invention also proposes an electronic device, please refer to [link to relevant documentation]. Figure 3The image shows an electronic device according to Embodiment 3 of the present invention, including a memory 20, a processor 10, and a computer program 30 stored in the memory and executable on the processor. When the processor 10 executes the computer program 30, it implements the hepatocellular carcinoma data processing method described above.

[0107] In some embodiments, the processor 10 may be a central processing unit (CPU), controller, microcontroller, microprocessor or other data processing chip, used to run program code stored in memory 20 or process data, such as executing access restriction programs.

[0108] The memory 20 includes at least one type of readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 20 can be an internal storage unit of an electronic device, such as the hard disk of the electronic device. In other embodiments, the memory 20 can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, the memory 20 can include both internal and external storage units of the electronic device. The memory 20 can be used not only to store application software and various types of data of the electronic device, but also to temporarily store data that has been output or will be output.

[0109] It should be pointed out that, Figure 3 The structure shown does not constitute a limitation on the electronic device. In other embodiments, the electronic device may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0110] This invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the hepatocellular carcinoma data processing method described above.

[0111] Those skilled in the art will understand that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0112] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0113] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0114] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0115] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A method for processing hepatocellular carcinoma data, characterized in that, The method includes: Using the CIBERSORT tool combined with weighted gene co-expression network analysis, we screened gene sets that were positively correlated with M2 macrophage infiltration. Target genes with prognostic value were identified from the gene set through differential expression analysis and prognostic correlation analysis. A dataset containing the target gene is obtained, and a comprehensive machine learning algorithm is used to generate several corresponding prognostic prediction models in each dataset. Calculate the C-index of each prognostic prediction model under different datasets, calculate the average C-index, and determine the prognostic prediction model corresponding to the maximum average C-index. Determine whether the prognostic prediction model corresponding to the maximum average value of the C-index is unique; If so, the prognostic prediction model corresponding to the maximum average value of C-index is determined as the target prognostic prediction model; If not, then based on the C-index of each prognostic prediction model under different datasets, calculate the complexity of the prognostic prediction model corresponding to the maximum average C-index, and determine the prognostic prediction model corresponding to the minimum complexity as the target prognostic prediction model. The step of calculating the complexity of the prognostic prediction model corresponding to the maximum average C-index of each prognostic prediction model under different datasets includes: Determine the likelihood function of the prognostic prediction model corresponding to the maximum average value of the C-index. The likelihood function includes the parameter vector of the prognostic prediction model. Take the logarithm of the likelihood function to obtain the log-likelihood function. Then, take the derivative with respect to the parameter vector and set the derivative to zero to find the parameter estimate that maximizes the log-likelihood function, thus obtaining the maximum likelihood estimate. The number of parameters in the prognostic prediction model corresponding to the maximum average value of the C-index is determined, and the complexity is calculated based on the number of parameters and the maximum likelihood estimate. The C-index of each prognostic prediction model under different datasets is ranked from largest to smallest. Based on the ranking results and the importance of the datasets, the adjustment factor corresponding to the C-index of each prognostic prediction model under different datasets is determined. The target complexity is calculated based on the adjustment factor and the complexity.

2. The hepatocellular carcinoma data processing method according to claim 1, characterized in that, The datasets are RNA-seq datasets from hepatocellular carcinoma users, including the TCGA dataset, ICGC dataset, GSE14520 dataset, and GSE76427 dataset.

3. The hepatocellular carcinoma data processing method according to claim 2, characterized in that, The steps of using the CIBERSORT tool combined with weighted gene co-expression network analysis to screen for gene sets positively correlated with M2 macrophage infiltration include: The overall RNA sequencing data from the TCGA dataset was input into the CIBERSORT tool to generate an immune cell infiltration expression matrix for each hepatocellular carcinoma sample in the corresponding dataset. We used weighted gene co-expression network analysis to analyze the immune cell infiltration expression matrix of each hepatocellular carcinoma sample in the dataset; The neighborhood relationships were converted into topological overlap matrices, and the average values ​​of different topological overlap matrix measures were used to cluster them into chain hierarchical structures. Gene sets that are positively correlated with M2 macrophage infiltration were then identified.

4. The hepatocellular carcinoma data processing method according to claim 3, characterized in that, In the step of analyzing the immune cell infiltration expression matrix of each hepatocellular carcinoma sample in the dataset using weighted gene co-expression network analysis, the independence β value was set to 0.9, and hepatocellular carcinoma samples with an average expression level >0.5 were selected for analysis.

5. The hepatocellular carcinoma data processing method according to claim 4, characterized in that, The step of identifying target genes with prognostic value from the gene set through differential expression analysis and prognostic correlation analysis includes: Transcriptome expression differential analysis was used to identify the first differentially expressed gene between hepatocellular carcinoma tissue and normal liver tissue in the TCGA dataset, with |LogFC|≥1.5 as the threshold. Here, |LogFC| is the absolute value of the logarithmic fold change. Identify the second differentially expressed gene between hepatocellular carcinoma tissue and normal liver tissue by finding the same members of the first differentially expressed gene and M2 macrophages. Cox regression analysis was used to compare the correlation between the expression of the second differentially expressed gene and the overall survival time of hepatocellular carcinoma patients to identify the target gene.

6. The hepatocellular carcinoma data processing method according to claim 5, characterized in that, The comprehensive machine learning algorithms include eight individual algorithms and 22 joint algorithms, including SurvivalSVM, SuperPC, Enet, StepCox, Ridge Regression, plsRcox, RSF, and GBM.

7. A hepatocellular carcinoma data processing system, characterized in that, The system for implementing the hepatocellular carcinoma data processing method according to any one of claims 1-6, the system comprising: The screening module is used to use the CIBERSORT tool combined with weighted gene co-expression network analysis to screen gene sets that are positively correlated with M2 macrophage infiltration; The first determination module is used to determine target genes with prognostic value from the gene set through differential expression analysis and prognostic correlation analysis. The generation module is used to acquire a dataset containing the target gene and to generate several corresponding prognostic prediction models in each dataset using a comprehensive machine learning algorithm. The calculation module is used to calculate the C-index of each prognostic prediction model under different datasets, calculate the average value of the C-index, and determine the prognostic prediction model corresponding to the maximum average value of the C-index. The judgment module is used to determine whether the prognostic prediction model corresponding to the maximum average value of C-index is unique. The second determining module is used to determine the prognostic prediction model corresponding to the maximum average value of C-index as the target prognostic prediction model if the prognostic prediction model corresponding to the maximum average value of C-index is unique. The third determination module is used to calculate the complexity of the prognostic prediction model corresponding to the maximum average value of the C-index based on the C-index of each prognostic prediction model under different datasets if the prognostic prediction model corresponding to the maximum average value of the C-index is not unique, and determine the prognostic prediction model corresponding to the minimum complexity value as the target prognostic prediction model.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the hepatocellular carcinoma data processing method as described in any one of claims 1-6.

9. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the program, implements the hepatocellular carcinoma data processing method as described in any one of claims 1-6.

Images