Method and system for constructing autophagy-based acute myelogenous leukemia prognosis model
By constructing a prognosis model of acute myeloid leukemia based on autophagy genes, combining autophagy genes and clinical information, key prognostic markers were screened out, and the problem of inaccurate prognosis in traditional prediction methods was solved, and the identification and clinical identification of high-risk patients were achieved. Optimization of management.
Patent Information
- Application Number
- CN202411586134.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-08
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional methods for prognostic prediction of acute myeloid leukemia ignore the importance of autophagy and immune microenvironment, resulting in insufficient accuracy of prognostic markers, and the use of single algorithms often leads to error generation.
By obtaining autophagy gene data, building an autophagy gene set, combining clinical information and transcriptome data, using log-rank segmentation rules, support vectors and weight coefficient vectors, key autophagy gene sets were screened out for regression analysis and Survival analysis, a prognostic model of acute myeloid leukemia based on autophagy was constructed, and prognostic markers such as BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1 were identified.
It significantly improves the accuracy and robustness of the prognostic model, can identify high-risk acute myeloid leukemia patients, and optimizes clinical management.
Smart Images

Figure CN119091959A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of bioinformatics, and in particular relates to a method and system for constructing an acute myeloid leukemia prognosis model based on autophagy. Background Art
[0002] Acute myeloid leukemia (AML) is a complex and diverse blood cancer caused by abnormal proliferation and immature differentiation of hematopoietic stem cells in the bone marrow. Although the treatment of AML has improved, its efficacy is still not ideal.
[0003] Traditional prognosis prediction methods usually rely mainly on clinical and molecular marker data, ignoring the importance of autophagy and immune microenvironment. In addition, when screening prognostic genes, a single algorithm is often used, which often produces certain errors due to the limitations of the algorithm, resulting in inaccurate prognostic markers. Summary of the invention
[0004] In order to solve the problems raised by the background technology, the present invention provides a method and system for constructing an acute myeloid leukemia prognosis model based on autophagy.
[0005] The technical solution of the present invention is as follows: The present invention provides a method for constructing an acute myeloid leukemia prognosis model based on autophagy, comprising the following steps: S1: Obtain autophagy gene data and construct autophagy gene set; Obtain clinical information data and transcriptome data, combine the clinical information data with the transcriptome data to obtain the sample data to be tested; extract the autophagy genes in the transcriptome data according to the autophagy gene set to obtain the target autophagy gene set; S2: Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels. A decision tree is constructed using the log-rank segmentation rule. After the variables are segmented, the importance scores of the variables are calculated to obtain the first key autophagy gene set. Based on the target autophagy gene set, the importance score of each target autophagy gene is calculated according to the support vector and the weight coefficient vector to obtain the second key autophagy gene set; Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels, and the gain value of the variable is calculated to obtain the third key autophagy gene set; The intersection of the first key autophagy gene set, the second key autophagy gene set and the third key autophagy gene set is taken to obtain the key autophagy gene set; S3: The key autophagy gene set was subjected to regression analysis and survival analysis to obtain the candidate gene set, and the candidate gene set was subjected to stepwise regression analysis to obtain the prognostic markers and regression coefficients; The prognostic markers are BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1; S4: Based on prognostic markers and regression coefficients, according to the formula: , construct an autophagy-based prognostic model for acute myeloid leukemia; In the formula, Score the risk; , , , , , They are the expression levels of prognostic markers BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1, respectively.
[0006] In step S2, the first key autophagy gene set is obtained, specifically: The target autophagy gene set was used as a variable, and the survival time and survival status in the sample data to be tested were used as labels. A decision tree was constructed using the log-rank segmentation rule. After the variables were segmented, for each candidate split point, the statistics of the two subsets after the split were calculated. The importance score of the variable was calculated based on the statistics, and the variable with an importance score greater than the preset importance score was taken as the first key autophagy gene to obtain the first key autophagy gene set.
[0007] In step S2, the second key autophagy gene set is obtained, specifically: According to the support vector and the weight coefficient vector, the absolute value of the weight coefficient of each target autophagy gene is calculated to obtain the importance score; the importance score is converted into a vector and mapped to the corresponding target autophagy gene, and the target autophagy gene with an importance score greater than the preset importance score is taken as the second key autophagy gene to obtain the second key autophagy gene set.
[0008] In step S2, the third key autophagy gene set is obtained, specifically: The target autophagy gene set was used as a variable, and the survival time and survival status in the sample data to be tested were used as labels. The objective function was constructed, and the Cox proportional hazard model was trained. According to the trained Cox proportional hazard model, the gain value of the variable was calculated, and the variable with a gain value greater than the preset gain value was taken as the third key autophagy gene to obtain the third key autophagy gene set.
[0009] The present invention also provides a system for constructing an acute myeloid leukemia prognosis model based on autophagy, comprising: Target autophagy gene set construction module: used to obtain autophagy gene data and construct autophagy gene sets; Obtain clinical information data and transcriptome data, combine the clinical information data with the transcriptome data to obtain the sample data to be tested; extract the autophagy genes in the transcriptome data according to the autophagy gene set to obtain the target autophagy gene set; Key autophagy gene set construction module: Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels. A decision tree is constructed through the log-rank segmentation rule. After the variables are segmented, the importance score of the variables is calculated to obtain the first key autophagy gene set; Based on the target autophagy gene set, the importance score of each target autophagy gene is calculated according to the support vector and the weight coefficient vector to obtain the second key autophagy gene set; Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels, and the gain value of the variable is calculated to obtain the third key autophagy gene set; The intersection of the first key autophagy gene set, the second key autophagy gene set and the third key autophagy gene set is taken to obtain the key autophagy gene set; Prognostic marker acquisition module: The key autophagy gene set is subjected to regression analysis and survival analysis to obtain the candidate gene set, and the candidate gene set is subjected to stepwise regression analysis to obtain the prognostic markers and regression coefficients; The prognostic markers are BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1; Prognostic model building module: Based on prognostic markers and regression coefficients, according to the formula: , construct an autophagy-based prognostic model for acute myeloid leukemia; In the formula, Score the risk; , , , , , They are the expression levels of prognostic markers BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1, respectively.
[0010] Beneficial effects: The present invention incorporates autophagy genes into the analysis, studies the effect of autophagy characteristics on the prognosis of acute myeloid leukemia, uses different processing methods combined with regression analysis and survival analysis, screens out potential autophagy gene prognostic markers, constructs a prognostic model based on autophagy characteristics, identifies high-risk acute myeloid leukemia patients, and significantly improves the accuracy and robustness of the prognostic model, thereby optimizing clinical management. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] Figure 1Schematic diagram of a process for constructing an acute myeloid leukemia prognosis model based on autophagy in an embodiment.
[0012] Figure 2 This is the Out-of-Bag (OOB) error graph of the first model training in the embodiment.
[0013] Figure 3 Variable Importance (VIMP) diagram of the importance score of the target autophagy gene in the embodiment.
[0014] Figure 4 4 is a comparison chart of the number of iterations of the model training process and the Cox negative log-likelihood value in the embodiment.
[0015] Figure 5 4 is a Venn diagram of overlapping genes of the first key autophagy gene set, the second key autophagy gene set and the third key autophagy gene set in the embodiment.
[0016] Figure 6 It is the path diagram of the LASSO coefficient in the embodiment.
[0017] Figure 7 It is the cross validation curve of LASSO regression analysis in the embodiment.
[0018] Figure 8 : It is the risk score distribution, survival status and gene expression heat map of 6 prognostic markers of the training set in the embodiment.
[0019] Fig. 9 It is the Kaplan-Meier curve of the training set in the embodiment.
[0020] Fig.10 : This is the AUC curve of the acute myeloid leukemia prognosis model based on autophagy for the training set in the embodiment.
[0021] Fig.11 (a) is a single factor Cox forest plot in the embodiment, and (b) is a multi-factor Cox forest plot.
[0022] Fig.12 : is a Norman plot of the comprehensive risk score and other survival information in the embodiment.
[0023] Fig.13 (a) is a calibration curve for predicting the survival rate of acute myeloid leukemia patients in the training set at 365 days (1 year), (b) is 1095 days (3 years), and (c) is 1825 days (5 years) in the embodiment.
[0024] Fig.14The results of the validation set of the autophagy-based acute myeloid leukemia prognostic model in the embodiment, wherein (A) is the Kaplan-Meier curve; (B) is the AUC curve; and (C) is the risk score distribution, survival status, and gene expression heat map of 6 prognostic markers.
[0025] Fig.15 : is a heat map of the correlation of the abundance of different immune cells in the embodiment.
[0026] Fig.16 It is a violin plot of the abundance of immune cells in the embodiment.
[0027] Fig.17 This is a heat map of the correlation between the 6 prognostic markers and immune cells in the example. DETAILED DESCRIPTION
[0028] The following examples are intended to illustrate the present invention rather than to further limit the present invention.
[0029] The present invention provides a method for constructing an acute myeloid leukemia prognosis model based on autophagy, such as Figure 1 As shown, the following steps are included: S1: Obtain autophagy gene data and construct autophagy gene set; Clinical information data and transcriptome data are obtained, and the clinical information data and transcriptome data are combined to obtain the sample data to be tested; according to the autophagy gene set, the autophagy genes in the transcriptome data are extracted to obtain the target autophagy gene set.
[0030] The specific operations are as follows: Autophagy gene data were obtained from the human autophagy gene database (HADB) and the GO_AUTOPHAGY gene set of the GSEA website, respectively. The two parts of autophagy gene data were merged to construct an autophagy gene set, and 531 related autophagy genes (ARGs) were obtained.
[0031] The GSE37642 original microarray dataset was downloaded from the Gene Expression Omnibus (GEO) database, and its transcriptome data from the GPL96 platform and GPL570 platform were obtained. The transcriptome data from the two platforms were merged and the batch effect was processed using the sva package. Then clinical information data was obtained, and the processed transcriptome data was combined with the clinical information data. After removing samples lacking clinical information, a total of 553 acute myeloid leukemia (AML) samples were obtained as the training set.
[0032] Finally, according to the autophagy gene set, that is, from the 531 related ARGs, we screened the autophagy genes present in the transcriptome data of the training set and obtained 392 ARGs as the target autophagy gene set.
[0033] S2: Based on the target autophagy gene set, the target autophagy gene set is used as a variable, the survival time and survival status in the sample data to be tested are used as labels, and a decision tree is constructed using the log-rank segmentation rule. After the variables are segmented, the importance score of the variables is calculated to obtain the first key autophagy gene set. Specifically: The target autophagy gene set was used as a variable, and the survival time and survival status in the sample data to be tested were used as labels. A decision tree was constructed using the log-rank segmentation rule. After the variables were segmented, for each candidate split point, the statistics of the two subsets after the split were calculated. The importance score of the variable was calculated based on the statistics, and the variable with an importance score greater than the preset importance score was taken as the first key autophagy gene to obtain the first key autophagy gene set.
[0034] The above processing method was used to perform survival analysis on 392 ARGs. The specific operations are as follows: The training set contains multiple samples, each of which has multiple target autophagy genes and their corresponding survival information. The survival time and survival status information are extracted from the training set. In order to predict the survival rate of the sample, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels. The first model containing 1000 decision trees is constructed by the log-rank segmentation rule. The variables are segmented by the log-rank segmentation rule. For each candidate split point, the statistics of the two subsets after the split are calculated. The split point with the largest log-rank statistic is selected as the optimal split point. The variable data is segmented according to the optimal split point until the first model stopping condition is met. Figure 2 As shown in the figure, as the number of trees increases, the error of the first model gradually decreases and the generalization performance improves. The importance score of the variable is calculated based on the statistic. The importance score represents the importance of each target autophagy gene and is used to identify genes that have a great impact on the prediction of the first model. Variables with an importance score greater than the preset importance score are taken as the first key autophagy gene, and 146 autophagy genes that have a significant impact on survival analysis are screened out and recorded as the first key autophagy gene set.
[0035] Based on the target autophagy gene set, the importance score of each target autophagy gene is calculated according to the support vector and weight coefficient vector to obtain the second key autophagy gene set. Specifically: According to the support vector and the weight coefficient vector, the absolute value of the weight coefficient of each target autophagy gene is calculated to obtain the importance score; the importance score is converted into a vector and mapped to the corresponding target autophagy gene, and the target autophagy gene with an importance score greater than the preset importance score is taken as the second key autophagy gene to obtain the second key autophagy gene set.
[0036] The above processing method was used to perform survival analysis on 392 ARGs. The specific operations are as follows: Using the support vector and weight coefficient vector, the importance score of each target autophagy gene is calculated as follows: , In the formula, is the importance score, is the weight coefficient vector, is the support vector.
[0037] The calculated importance scores are converted into vector form and mapped to the corresponding target autophagy genes, that is, each target autophagy gene is assigned a corresponding importance score. The feature importance scores are then sorted in descending order, such as Figure 3 As shown, the target autophagy genes with importance scores greater than the preset importance scores were taken as the second key autophagy genes, and the top 180 autophagy genes that had a significant impact on survival prediction were filtered out and recorded as the second key autophagy gene set.
[0038] Based on the target autophagy gene set, the target autophagy gene set is used as a variable, the survival time and survival status in the sample data to be tested are used as labels, and the gain value of the variable is calculated to obtain the third key autophagy gene set. Specifically: Taking the target autophagy gene set as variables and the survival time and survival status in the sample data to be tested as labels, an objective function is constructed, and a processing model is trained. According to the trained processing model, the gain value of the variable is calculated, and the variable with a gain value greater than the preset gain value is taken as the third key autophagy gene to obtain the third key autophagy gene set.
[0039] The above processing method was used to perform survival analysis on 392 ARGs. The specific operations are as follows: First, the target autophagy gene set is used as a variable, and then the training set data is preprocessed. The survival time and survival status are labeled and converted into DMatrix format. The processing model parameters are set. The objective function is the Cox proportional hazard model, and the evaluation index is the Cox negative log likelihood (Cox-nloglik). During the training process, the number of iterations can be set to 100 times, and the learning rate can be set to 0.1. The processing model gradually converges during the training process, and the Cox negative log likelihood value continues to decrease, indicating that the prediction performance of the processing model gradually improves, such as Figure 4 As shown. Then, the trained processing model was used to calculate the gain value of the variable, which was used to evaluate the importance of the target autophagy gene. The importance indicated the contribution of each target autophagy gene to the processing model. Finally, the variable with a gain value greater than the preset gain value was taken as the third key autophagy gene, and 180 autophagy genes with significant effects on survival analysis were screened out and recorded as the third key autophagy gene set.
[0040] The intersection of the first key autophagy gene set (146), the second key autophagy gene set (180), and the third key autophagy gene set (180) was taken, such as Figure 5 As shown, a total of 45 autophagy genes were obtained, which were recorded as the key autophagy gene set.
[0041] S3: The key autophagy gene set was subjected to regression analysis and survival analysis to obtain the candidate gene set, and the candidate gene set was subjected to stepwise regression analysis to obtain the prognostic markers and regression coefficients.
[0042] In order to avoid overfitting the prognostic risk characteristics, after the key autophagy genes were screened by three different treatment methods, regression analysis and survival analysis, such as Lasso-Cox analysis, were used to further screen the autophagy genes significantly associated with survival and conduct modeling. The specific operations are as follows: First, the optimal lambda value (lambda.1se) is selected as 0.09393562 through 10-fold cross validation, and the path diagram of the LASSO coefficient and the cross-validation curve of the LASSO regression analysis are drawn, as shown in Figure 6 , Figure 7 Then extract the non-zero coefficients and save the candidate genes and their regression coefficients selected by the regression analysis.
[0043] Then, a stepwise regression analysis, such as Cox stepwise regression analysis, is performed to optimize the selection of characteristic genes and obtain prognostic markers.
[0044] Preferably, the prognostic markers are BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1.
[0045] S4: Based on prognostic markers and regression coefficients, an autophagy-based prognostic model for acute myeloid leukemia was constructed. Specifically: After calculating the product of the expression level of each prognostic marker and the corresponding regression coefficient, the score of each prognostic marker was obtained, and the scores of all prognostic markers were added together to obtain the risk score. The survival status was predicted according to the risk score, and an autophagy-based acute myeloid leukemia prognostic model was constructed.
[0046] The autophagy-based prognostic model for acute myeloid leukemia is as follows: , In the formula, Score the risk; , , , , , They are the expression levels of prognostic markers BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1, respectively.
[0047] The present invention incorporates autophagy genes into the analysis, studies the effect of autophagy characteristics on the prognosis of acute myeloid leukemia, uses different processing methods combined with regression analysis and survival analysis to screen out potential autophagy gene prognostic markers, constructs a prognostic model based on autophagy characteristics, identifies high-risk AML patients, and significantly improves the accuracy and robustness of the prognostic model, thereby optimizing clinical management.
[0048] Experimental Results 1. Testing the predictive performance of the prognostic model like Figure 8 As shown, using the prognostic model of the present application, the risk score of each sample in the training set was calculated, and the samples were divided into high-risk group and low-risk group according to the risk score cutoff threshold. An increase in the risk score was associated with an increase in the number of deaths in patients. Among the prognostic markers, DAPK1, UBQLN4 and BAG3 were highly expressed in high-risk, and ULK1, ALCOCO2 and TSC2 were highly expressed in low-risk.
[0049] The prognostic model of the present application was used to predict the overall survival of the training set samples, wherein, in order to evaluate the difference in survival time, the Kaplan-Meier survival curve was used, such as Fig. 9 The results showed that the overall survival (OS) of patients in the high-risk group was shorter than that in the low-risk group (P<0.0001). Further, the accuracy of the constructed prognostic model was evaluated. Fig.10 ,The results showed that the AUC (area under the ROC curve) values of 1-year, 3-year, and 5-year OS were 0.660, 0.733, and 0.739, respectively, indicating that the prognostic model of the present application has a high predictive ability.
[0050] Univariate Cox regression analysis and multivariate Cox regression analysis were used to verify the independence of prognosis-related autophagy genes in predicting survival. Fig.11 Univariate Cox regression analysis in (a) showed that age, runx1 mutation, and risk score were significantly associated with OS of patients. Fig.11 Multivariate Cox regression analysis in (b) showed that age and risk score were independent predictors of AML patients.
[0051] In order to more accurately evaluate the effectiveness of the survival prediction model, a Norman plot of the comprehensive risk score and other survival information was constructed, e.g. Fig.12 . Further, Fig.13 The calibration curves in (a), (b), and (c) can accurately predict the OS of AML patients, which indicates that combining the risk score obtained by the prognostic model of the present application with clinical information can enhance the prediction of OS.
[0052] In order to further verify the predictive ability of the prognostic model, the dataset GSE12417 was processed in the same way as GSE37642 to obtain 237 samples with clinical information as the validation set. Fig.14 Among them, Fig.14 In Figure A, the comparison of overall survival (OS) using Kaplan-Meier survival curve and log-rank test showed that the OS of patients in the high-risk group was significantly shorter than that of patients in the low-risk group (P<0.0001). In order to further evaluate the classification performance of the prognostic model for patient survival in different data sets, the ROC curve of patient survival was drawn according to the model risk score, as shown in Figure 2. Fig.14 In the validation set, the AUC values of 1-year and 3-year OS were 0.633 and 0.651, respectively, indicating that the prognostic model of this application has strong predictive ability. In addition, this study also analyzed the risk score and OS distribution of patients, such as Fig.14 In C, it was found that the mortality rate in the high-risk group was higher than that in the low-risk group. In terms of gene expression, DAPK1, UBQLN4, and BAG3 in the high-risk group were significantly upregulated, and ULK1, ALCOCO2, and TSC2 in the low-risk group were significantly downregulated, which was consistent with the risk score calculation results. Therefore, the validation results show that the prognostic model of this application has reasonable accuracy and the ability to independently predict the OS of AML patients.
[0053] 2. Differential expression analysis, functional enrichment analysis and immune infiltration analysis The limma package was used to perform differential expression analysis on the transcriptome data of the high-risk group and low-risk group samples in the training set, and the differentially expressed genes were screened with |log2FC|>2 and p value <0.05 as the criteria. Finally, 63 differentially expressed genes were identified, including 47 up-regulated genes and 16 down-regulated genes.
[0054] ClusterProfiler (version 3.14.3) was used to analyze the differentially expressed genes according to gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) to reveal the main functions of the differentially expressed genes. The Benjamini&Hochberg correction method was used, and the corrected P value less than 0.05 was used as the standard for statistical significance. Functional enrichment analysis showed that the differentially expressed genes were mainly related to biological processes such as thymic T cell differentiation and lymphocyte differentiation. In terms of cellular components, the differentially expressed genes were mainly present in the lumen of tertiary granules, actin filament bundles, and platelet α granules, and were involved in molecular functions such as chemokine activity and cytokine receptor binding. KEGG pathway analysis showed that the differentially expressed genes were mainly enriched in the IL-17 signaling pathway and Th1 and Th2 cell differentiation. The above analysis results indicate that the differentially expressed genes may play a role in the prognosis and immune response of AML.
[0055] Furthermore, immune infiltration analysis was performed on autophagy genes. The CIBERSORT algorithm was used to calculate the infiltration abundance of 22 immune cell types in the gene expression data of AML patients. The “ggcorrplot” package was used to detect the correlation between autophagy genes and the abundance of 22 immune cells and visualize them, as shown in Figure 2. Fig.15 shown. Fig.16 The immune infiltration results showed that the abundance of immune cells in the low-risk group of AML patients, including activated memory CD4 + T cells, activated NK cells, and resting memory CD4 + T cells, which were higher than those in the high-risk group. In addition, the results also showed that the correlation between autophagy genes and immune cells was greater than 0.28.
[0056] In addition, the relationship between 6 prognostic markers and immune infiltration was also explored, such as Fig.17 The results showed that these six prognostic markers were associated with resting memory CD4 + T cells, CD8 + T cells, M1 macrophages and immune cells are related, indicating that changes in the abundance of these immune cells may affect the pathogenesis of AML. The above results indicate that the prognostic markers of the present application may affect the abundance of immune cells in AML patients, thereby weakening the immune system's control over acute myeloid leukemia, thereby affecting the survival of acute myeloid leukemia.
[0057] The present invention divides acute myeloid leukemia into high-risk and low-risk groups according to the prognostic model, performs differential expression analysis, and then performs functional enrichment analysis on the genes with significant differences. The results show that these autophagy genes are mainly enriched in immune-related pathways such as thymic T cell differentiation and lymphocyte differentiation. Based on this, immune infiltration analysis of autophagy genes was performed. In addition, the relationship between prognostic markers and immune infiltration was studied, and the prognostic model and the effect of a single prognostic marker on immune abundance were analyzed, thereby revealing the relationship between autophagy genes and immune cells and the survival prognosis of acute myeloid leukemia, which helps to gain a deeper understanding of the immune characteristics and regulatory mechanisms of the tumor microenvironment (TME), and provides new targets and strategies for cancer treatment.
[0058] The present invention also provides a system for constructing an acute myeloid leukemia prognosis model based on autophagy, comprising: Target autophagy gene set construction module: used to obtain autophagy gene data and construct autophagy gene sets; Obtain clinical information data and transcriptome data, combine the clinical information data with the transcriptome data to obtain the sample data to be tested; extract the autophagy genes in the transcriptome data according to the autophagy gene set to obtain the target autophagy gene set; Key autophagy gene set construction module: Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels. A decision tree is constructed through the log-rank segmentation rule. After the variables are segmented, the importance score of the variables is calculated to obtain the first key autophagy gene set; Based on the target autophagy gene set, the importance score of each target autophagy gene is calculated according to the support vector and the weight coefficient vector to obtain the second key autophagy gene set; Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels, and the gain value of the variable is calculated to obtain the third key autophagy gene set; The intersection of the first key autophagy gene set, the second key autophagy gene set and the third key autophagy gene set is taken to obtain the key autophagy gene set; Prognostic marker acquisition module: The key autophagy gene set is subjected to regression analysis and survival analysis to obtain the candidate gene set, and the candidate gene set is subjected to stepwise regression analysis to obtain the prognostic markers and regression coefficients; The prognostic markers are BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1; Prognostic model building module: Based on prognostic markers and regression coefficients, according to the formula: , construct an autophagy-based prognostic model for acute myeloid leukemia; In the formula, Score the risk; , , , , , They are the expression levels of prognostic markers BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1, respectively.
Claims
1. A method for constructing an acute myeloid leukemia prognosis model based on autophagy. It is characterized in that The following steps are involved: S1: Obtain autophagy gene data and construct autophagy gene set; Acquire clinical information data and transcriptome data, combine the clinical information data with the transcriptome data to obtain the sample data to be tested; According to the autophagy gene set, autophagy genes in the transcriptome data are extracted to obtain the target autophagy gene set; S2: Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels. A decision tree is constructed using the log-rank segmentation rule. After the variables are segmented, the importance scores of the variables are calculated to obtain the first key autophagy gene set. Based on the target autophagy gene set, the importance score of each target autophagy gene is calculated according to the support vector and the weight coefficient vector to obtain the second key autophagy gene set; Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels, and the gain value of the variable is calculated to obtain the third key autophagy gene set; The intersection of the first key autophagy gene set, the second key autophagy gene set and the third key autophagy gene set is taken to obtain the key autophagy gene set; S3: The key autophagy gene set was subjected to regression analysis and survival analysis to obtain the candidate gene set, and the candidate gene set was subjected to stepwise regression analysis to obtain the prognostic markers and regression coefficients; The prognostic markers are BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1; S4: Based on prognostic markers and regression coefficients, according to the formula: , construct an autophagy-based prognostic model for acute myeloid leukemia; In the formula, Score the risk; , , , , , They are the expression levels of prognostic markers BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1, respectively.
2. The method for constructing an autophagy-based acute myeloid leukemia prognosis model according to claim 1, It is characterized in that In step S2, the first key autophagy gene set is obtained, specifically: The target autophagy gene set was used as the variable, and the survival time and survival status in the sample data to be tested were used as labels. A decision tree was constructed using the log-rank segmentation rule. After the variables were segmented, for each candidate split point, the statistics of the two subsets after the split were calculated. The importance score of the variable was calculated based on the statistics, and the variable with an importance score greater than the preset importance score was taken as the first key autophagy gene to obtain the first key autophagy gene set.
3. The method for constructing an autophagy-based acute myeloid leukemia prognosis model according to claim 1, It is characterized in that In step S2, the second key autophagy gene set is obtained, specifically: According to the support vector and weight coefficient vector, the absolute value of the weight coefficient of each target autophagy gene is calculated to obtain the importance score; The importance score is converted into a vector and mapped to the corresponding target autophagy gene. The target autophagy gene with an importance score greater than the preset importance score is taken as the second key autophagy gene to obtain the second key autophagy gene set.
4. The method for constructing an autophagy-based acute myeloid leukemia prognosis model according to claim 1, It is characterized in that In step S2, the third key autophagy gene set is obtained, specifically: The target autophagy gene set was used as a variable, and the survival time and survival status in the sample data to be tested were used as labels. The objective function was constructed, and the Cox proportional hazard model was trained. According to the trained Cox proportional hazard model, the gain value of the variable was calculated, and the variable with a gain value greater than the preset gain value was taken as the third key autophagy gene to obtain the third key autophagy gene set.
5. A system for constructing an acute myeloid leukemia prognosis model based on autophagy. It is characterized in that include: Target autophagy gene set construction module: used to obtain autophagy gene data and construct autophagy gene sets; Acquire clinical information data and transcriptome data, combine the clinical information data with the transcriptome data to obtain the sample data to be tested; According to the autophagy gene set, autophagy genes in the transcriptome data are extracted to obtain the target autophagy gene set; Key autophagy gene set construction module: Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels. A decision tree is constructed through the log-rank segmentation rule. After the variables are segmented, the importance score of the variables is calculated to obtain the first key autophagy gene set; Based on the target autophagy gene set, the importance score of each target autophagy gene is calculated according to the support vector and the weight coefficient vector to obtain the second key autophagy gene set; Based on the target autophagy gene set, the target autophagy gene set is used as a variable, and the survival time and survival status in the sample data to be tested are used as labels, and the gain value of the variable is calculated to obtain the third key autophagy gene set; The intersection of the first key autophagy gene set, the second key autophagy gene set and the third key autophagy gene set is taken to obtain the key autophagy gene set; Prognostic marker acquisition module: The key autophagy gene set is subjected to regression analysis and survival analysis to obtain the candidate gene set, and the candidate gene set is subjected to stepwise regression analysis to obtain the prognostic markers and regression coefficients; The prognostic markers are BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1; Prognostic model building module: Based on prognostic markers and regression coefficients, according to the formula: , construct an autophagy-based prognostic model for acute myeloid leukemia; In the formula, Score the risk; , , , , , They are the expression levels of prognostic markers BAG3, TSC2, CALCOCO2, UBQLN4, ULK1, and DAPK1, respectively.
Citation Information
Patent Citations
Angiogenesis-related gene-based tumor prognosis prediction model construction method
CN114863999A
Prognosis model for predicting laryngeal squamous cell carcinoma based on autophagy-related genes and construction method thereof
CN115631797A
Prognosis model of hereditary ovarian cancer autophagy gene and construction method thereof
CN117144009A
Construction method and system of acute myelogenous leukemia prognosis model
CN118841180A
