Prognostic risk prediction method and equipment based on three-level lymphatic structure characteristic gene

Abstract

The invention discloses a prognosis risk prediction method based on a three-level lymphatic structure characteristic gene. The method comprises the following steps: dividing a transcription map of a GBM sample of a training data set into different GBM molecular subtypes; differentially expressed genes in different GBM molecular subtypes are identified, differentially expressed genes with prognosis values are screened, TLS related characteristic genes related to prognosis are obtained, and prognosis risk scores of all GBM samples of the training data set are calculated; and dividing each to-be-predicted GBM sample of the to-be-predicted GBM sample set into a high-risk score group and a low-risk score group according to the risk score optimal threshold. The invention further discloses computer equipment, a computer readable storage medium and a computer program product. The blank that classification indexes of immune-related molecular spectrums are lacked in the existing GBM molecular typing standard is effectively filled up; the bottleneck problem of effectively predicting the prognosis means of the GBM subject is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of prognosis prediction, and in particular relates to a prognosis prediction method and device based on characteristic genes related to tertiary lymphatic structures. Background Art

[0002] The 5-year mortality rate of brain glioma is second only to pancreatic cancer and lung cancer among systemic tumors, and it is the leading cause of death from neoplastic diseases in young and middle-aged men. Glioblastoma (GBM) is the most common and most malignant brain glioma, characterized by high disability rate and high recurrence rate, with a 5-year survival rate of about 6.6%, which seriously threatens human health. Immunotherapy has led to disruptive changes in the field of cancer treatment and has achieved remarkable therapeutic effects in GBM-related clinical trials. Although the 5th edition of the WHO Central Nervous System Tumors Classification, newly promulgated in 2021, integrates and optimizes the molecular typing standards of GBM, laying a solid foundation for the precise diagnosis and treatment of GBM, it does not include immune-related molecular spectrum characteristics, and its key guiding role in clinical immunotherapy decision-making is very limited. Therefore. There is an urgent need to construct a new method that can accurately identify GBM immune-related molecular subtypes, stratify and screen subjects with different prognostic risks, and then effectively predict the prognostic risks of subjects.

[0003] Tertiary lymphoid structure (TLS) is an organized ectopic lymphoid structure formed by the aggregation of lymphocytes in non-lymph node organs under pathological and physiological conditions. It exists in multiple tumor types including GBM and is an important site for the interaction between tumor cells and immune cells. It is closely related to the anti-tumor immune response and prognosis of subjects. TLS-related characteristic genes are important factors in predicting the efficacy of immunotherapy and the prognosis of subjects. However, there is currently a lack of a risk model for effectively predicting the prognosis of GBM subjects based on TLS-related characteristic genes.

[0004] The prognostic risk prediction model constructed by the present invention will effectively fill the gap in the classification indicators of the lack of immune-related molecular spectra in the current GBM molecular typing standards; and solve the bottleneck problem of effectively predicting the prognosis of GBM subjects. Summary of the invention

[0005] In view of the above problems existing in the prior art, the present invention provides a prognosis risk prediction method and device based on TLS-related characteristic genes.

[0006] The above-mentioned object of the present invention is achieved by the following technical solutions:

[0007] The prognostic risk prediction method based on the three-level lymphatic structure characteristic genes includes the following steps:

[0008] Step S11, based on the TLS-related gene feature set, the transcriptional profiles of the GBM samples in the training data set are divided into different GBM molecular subtypes by a consensus clustering algorithm;

[0009] Step S12, identifying differentially expressed genes in different GBM molecular subtypes, screening differentially expressed genes with prognostic value, obtaining TLS-related characteristic genes related to prognosis, and calculating the prognostic risk score of each GBM sample in the training data set;

[0010] Step S13: obtaining an optimal risk score threshold according to the prognostic risk score of each GBM sample in the training data set, and dividing each GBM sample to be predicted in the GBM sample set to be predicted into a high risk score group and a low risk score group according to the optimal risk score threshold.

[0011] As mentioned above, the GBM samples of the training data set are obtained from the TCGA-GBM training data set samples obtained from the Cancer Genome Atlas database.

[0012] As described above, differentially expressed genes in different GBM molecular subtypes were obtained based on the following steps:

[0013] Differentially expressed genes between different GBM molecular subtypes were screened through differential expression analysis to obtain differentially expressed genes;

[0014] The differentially expressed genes with prognostic value are obtained based on the following steps:

[0015] Through univariate proportional hazard Cox regression analysis, prognostic analysis was performed on all the differentially expressed genes, and multiple differentially expressed genes with prognostic value were screened;

[0016] The prognosis-related TLS-related characteristic genes are obtained based on the following steps:

[0017] The LASSO-Cox regression analysis algorithm was used to further screen the differentially expressed genes with prognostic value to obtain TLS-related characteristic genes related to prognosis.

[0018] As mentioned above, the prognostic risk score is obtained based on the following formula:

[0019]

[0020] Among them, expression i is the expression level of the i-th TLS-related characteristic gene related to prognosis in the GBM sample of the training data set, n is the number of TLS-related characteristic genes, coefficient i is the regression coefficient corresponding to the i-th TLS-related characteristic gene related to prognosis.

[0021] As described above, the optimal threshold for risk score is obtained based on the prognostic risk score of each GBM sample in the training dataset based on the following steps:

[0022] Based on the prognostic risk score value of each GBM sample in the training data set, the optimal threshold of the risk score for dividing the samples into high-risk group and low-risk group is calculated according to the statistical Youden index.

[0023] As described above, dividing each GBM sample to be predicted in the GBM sample set to be predicted into a high risk score group and a low risk score group according to the optimal risk score threshold includes the following steps:

[0024] Calculating the prognostic risk score of each GBM sample to be predicted in the GBM sample set to be predicted;

[0025] GBM samples to be predicted whose prognostic risk score is greater than the optimal risk score threshold are divided into the high-risk score group, and GBM samples to be predicted whose score is less than the optimal risk score threshold are divided into the low-risk score group.

[0026] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above prediction method when executing the computer program.

[0027] A computer-readable storage medium stores a computer program, which implements the steps of the above prediction method when executed by a processor.

[0028] A computer program product comprises a computer program, which implements the steps of the above prediction method when executed by a processor.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] Based on the TLS-related gene feature set, the present invention identifies two different GBM molecular subtypes in GBM samples, screens out a group of TLS-related characteristic genes that characterize GBM prognosis, and innovatively constructs a prognostic risk score that quantifies individual prognosis. Through verification in an independent validation data set, the prognostic risk score can be used as a biological indicator for predicting prognosis, which will effectively fill the gap in the current GBM molecular typing standard that lacks classification indicators for immune-related molecular spectra; and solve the bottleneck problem of effectively predicting the prognosis of GBM subjects. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 is a flow chart of the present invention;

[0032] Figure 2It is a consensus clustering result diagram based on the TLS-related gene feature set in the embodiment, consensus matrix is ​​a consensus matrix, k=2 means that the number of clusters is divided into 2 categories, 1 represents the first category GBM molecular subtype, and 2 represents the second category GBM molecular subtype;

[0033] Figure 3 Schematic diagram of LASSO-Cox regression analysis to screen TLS-related characteristic genes that characterize GBM prognosis. (A) Regression coefficient path diagram of differentially expressed genes with prognostic value, the ordinate represents the regression coefficient, and the abscissa represents the logarithm of the penalty coefficient; (B) Cross-validation diagram, the ordinate represents the minimum partial likelihood deviation value, the abscissa represents the logarithm of the penalty coefficient, and the top row numbers represent the number of independent variables (i.e., the number of TLS-related characteristic genes finally determined); (C) The regression coefficient result diagram of the five TLS-related characteristic genes finally obtained. The ordinate represents the gene name, and the abscissa represents the regression coefficient value.

[0034] Figure 4 The overall survival evaluation results of GBM samples in the training data set are shown in Figure 1. (A) Overall survival curves of high and low risk score groups. The ordinate represents the sample survival rate, the abscissa represents the number of survival days, the blue curve represents the low-risk group, the red curve represents the high-risk group, and the HR and 95% CI represent the optimal risk score and the corresponding 95% confidence interval. The numbers on the lower line represent the number of samples surviving at the corresponding time points in the two risk groups; (B) The efficiency results of risk score division into high and low risk groups. The ordinate represents the true positive rate, and the abscissa represents the false positive rate; the blue curve represents the area under the receiver operating curve for predicting 1-year overall survival, the red curve represents the area under the receiver operating curve for predicting 3-year overall survival, and the green curve represents the area under the receiver operating curve for predicting 5-year overall survival; (C) Sample risk score curve. The ordinate represents the risk score, the abscissa represents the number of samples, the blue curve represents the low-risk group, and the red curve represents the high-risk group; (D) Death and survival distribution of samples in high and low risk groups. The vertical axis represents the number of survival days, the horizontal axis represents the number of samples, the blue dots represent the samples in the low-risk group, and the red dots represent the samples in the high-risk group.

[0035] Figure 5Schematic diagram of the overall survival evaluation results of the GBM samples (CGGA.mRNAseq_325 samples) in the validation dataset. (A) Overall survival curves of the high and low risk score groups. The ordinate represents the sample survival rate, the abscissa represents the number of survival days, the blue curve represents the low-risk group, the red curve represents the high-risk group, and the HR and 95% CI represent the optimal risk score and the corresponding 95% confidence interval. The numbers on the lower line represent the number of samples surviving at the corresponding time points in the two risk groups; (B) The performance results of the risk score in predicting overall survival. The ordinate represents the true positive rate, and the abscissa represents the false positive rate; the blue curve represents the area under the receiver operating curve for predicting 1-year overall survival, the red curve represents the area under the receiver operating curve for predicting 3-year overall survival, and the green curve represents the area under the receiver operating curve for predicting 5-year overall survival. (C) Sample risk score curve. The ordinate represents the risk score, the abscissa represents the number of samples, the blue curve represents the low-risk group, and the red curve represents the high-risk group. (D) Death and survival distribution of samples in the high and low risk groups. The vertical axis represents the number of survival days, the horizontal axis represents the number of samples, the blue dots represent samples in the low-risk group, and the red dots represent samples in the high-risk group. DETAILED DESCRIPTION

[0036] In order to facilitate those skilled in the art to understand and implement the present invention, the present invention is further described in detail below in conjunction with embodiments. It should be understood that the implementation examples described herein are only used to illustrate and explain the present invention and are not used to limit the present invention.

[0037] In the following examples, the experimental methods used are conventional methods unless otherwise specified, and the materials, raw data, reagents, etc. used can be obtained from commercial channels or public databases unless otherwise specified.

[0038] Embodiment 1:

[0039] like Figure 1 As shown, the embodiment of the present invention provides a prognosis prediction method based on tertiary lymphoid structure (TLS) related characteristic genes, comprising the following specific steps:

[0040] Step S11: Based on the TLS-related gene feature set, the transcriptional profiles of the GBM samples in the training data set are divided into different GBM molecular subtypes by a consensus clustering algorithm. The TLS-related gene feature set can be a publicly published TLS-related gene feature set, or an authoritatively verified TLS-related gene feature set.

[0041] Optionally, the above step S11 may include the following steps S111-S113:

[0042] Step S111: data collection and preprocessing.

[0043] In a convenient and feasible embodiment of the present invention, the TCGA-GBM training data set samples are obtained from the Cancer Gene Atlas Database (https: / / portal.gdc.cancer.gov / ), and the CGGA.mRNAseq_325 validation data set samples are obtained from the Chinese Glioma Gene Atlas Database (http: / / www.cgga.org.cn / ). The TCGA-GBM cohort samples include 143 GBM samples with existing RNA sequencing data and clinical information as training data sets, and the CGGA-RNAseq325 cohort samples include 137 GBM samples with existing RNA sequencing data and clinical information as validation data sets. For high-throughput sequencing data, the FPKM expression profile (log transformation), survival information (overall survival) and clinical information (age, gender) of the TCGA-GBM cohort are downloaded.

[0044] The GBM samples in the TCGA-GBM training dataset and the CGGA.mRNAseq_325 validation dataset of the present invention all meet the following criteria: (1) confirmed as GBM by histology; (2) clinical overall survival information is available; (3) the number of GBM samples exceeds 100. The baseline statistical information of the above cohorts is shown in Tables 1 and 2. Table 1 is the baseline data of the TCGA-GBM training dataset samples; Table 2 is the baseline data of the CGGA-RNAseq325 validation dataset samples.

[0045] Table 1

[0046] Variable N N＝143 Age 143 Age<60 67(47％) Age>=60 76(53％) IDH.Mutation 140 Mut 10(7.1％) WT 130(93％) Unknown 3 Sex 143 Female 48(34％) Male 95(66％) MGMTp.Methylation 113 Methylated 46(41％) UnMethylated 67(59％) Unknown 30 1p19q.Codeletion 138 Non-codel 138(100％) Unknown 5

[0047] Table 2

[0048] Variable N N＝137 Age 137 Age<60 116(85％) Age>=60 21(15％) Sex 137 Female 50(36％) Male 87(64％)

[0049] Step S112, obtain a TLS-related gene feature set (i.e., a tertiary lymphoid structure gene feature set) consisting of 39 TLS-related genes from the literature published by Sautes-Fridman C et al. [Tertiary lymphoid structures in the era of cancer immunotherapy. Nature Reviews Cancer 2019, 19(6): 307-325.].

[0050] Step S113: using a consensus clustering algorithm, the transcriptional profiles of the GBM samples in the training data set are divided into different GBM molecular subtypes according to the acquired TLS-related gene feature set.

[0051] In a feasible embodiment of the present invention, based on the consensus clustering algorithm in the ConsensusClusterPlus package in the R package, the distance euclidean used for clustering, the clustering method Partitioning Around Medoids, and 1000 repetitions were performed to ensure clustering stability. The transcriptional profiles of GBM samples in the training data set were divided into different GBM molecular subtypes according to the TLS-related gene feature set. The overall survival differences of different GBM molecular subtypes were analyzed using Kaplan-Meier analysis and log-rank test.

[0052] The present invention adopts a consensus clustering algorithm and subdivides the transcriptional profiles of GBM samples into two different GBM molecular subtypes according to a TLS-related gene feature set consisting of 39 TLS-related genes, including 93 GBM samples in the first molecular subtype and 50 GBM samples in the second molecular subtype ( Figure 2 ).

[0053] Step S12: Identify differentially expressed genes in different GBM molecular subtypes, further screen differentially expressed genes with prognostic value, construct a prognostic risk scoring system, and evaluate the overall survival of the subjects.

[0054] Differential expression analysis screened out 212 differentially expressed genes between the two different molecular subtypes, and univariate proportional hazard Cox regression analysis further screened out differentially expressed genes associated with overall survival. A prognostic gene signature consisting of five TLS-related signature genes (PTGES, CXCL13, G0S2, PTX3, and PLAUR) was established using LASSO-Cox regression analysis. Figure 3 ). The prognostic risk score of the individual subjects was calculated based on the prognostic gene signature. According to the optimal threshold of the risk score, the subjects were divided into a high-risk score group and a low-risk score group in the training dataset TCGA-GBM cohort ( Figure 4 ).

[0055] Optionally, the above step S12 may include the following steps S121-S124:

[0056] Step S121, screening out differentially expressed genes between different GBM molecular subtypes through differential expression analysis to obtain differentially expressed genes.

[0057] In a feasible embodiment of the present invention, the differentially expressed genes between different GBM molecular subtypes are screened using the "limma" package, and the significance standard is |log2 Fold Change|>1 and the significance threshold FDR<0.05 to obtain the differentially expressed genes;

[0058] In this example, differential expression analysis screened out 212 differentially expressed genes between two different GBM molecular subtypes.

[0059] Step S122: Perform prognostic analysis on all the differentially expressed genes through univariate proportional hazard Cox regression analysis to screen out multiple differentially expressed genes with prognostic value.

[0060] In a feasible implementation of the present invention, 1000 randomizations are performed (each time accounting for 80% of all differentially expressed genes) to select a random differentially expressed gene set, the number of differentially expressed genes in the random differentially expressed gene set is 80% of all differentially expressed genes, for each random differentially expressed gene set, the prognostic value of the differentially expressed genes is further determined, and the statistical significance index P value of each differentially expressed gene is calculated using univariate proportional risk Cox regression analysis, and differentially expressed genes with P<0.05 are screened as differentially expressed genes with prognostic value.

[0061] In this example, 27 differentially expressed genes with prognostic value associated with overall survival were further screened out by univariate proportional hazard Cox regression analysis.

[0062] Step S123, further dimensionality reduction screening is performed on the plurality of differentially expressed genes with prognostic value, and the differentially expressed genes with prognostic value are further screened using the LASSO-Cox regression analysis algorithm to obtain TLS-related characteristic genes related to prognosis (R package glmnet).

[0063] In a feasible embodiment of the present invention, the multiple differentially expressed genes with prognostic value in the training data set are included as independent variables in LASSO-Cox regression analysis, and the dimensionality of the independent variables is reduced to finally obtain a group of TLS-related characteristic genes that characterize the prognosis of GBM samples.

[0064] The LASSO-Cox regression analysis algorithm is a compression estimation. It obtains a more refined analysis method by constructing a penalty function, which compresses the regression coefficients of some independent variables, that is, forces the sum of the absolute values ​​of the coefficients to be less than a fixed value, and sets the regression coefficients of some independent variables to zero. Therefore, the advantages of subset shrinkage are retained, and it is a biased estimation for processing data with complex collinearity.

[0065] In this embodiment, a group of TLS-related characteristic genes consisting of 5 prognosis-related TLS-related characteristic genes (PTGES, CXCL13, G0S2, PTX3, PLAUR) was established as a prognostic gene signature using the LASSO-Cox regression analysis algorithm.

[0066] Step S124: construct a prognostic risk scoring model based on the TLS-related characteristic genes related to prognosis obtained through screening, and evaluate the prognosis of GBM samples in the training data set.

[0067] The present invention calculates the prognostic risk score of each GBM sample in the training data set according to the expression levels of multiple prognosis-related TLS-related characteristic genes of each GBM sample in the training data set and the regression coefficients of the corresponding prognosis-related TLS-related characteristic genes obtained by LASSO-Cox regression analysis, and generates the following score calculation formula:

[0068]

[0069] Among them, expression i is the expression level of the i-th TLS-related characteristic gene related to prognosis in the GBM sample of the training data set, n is the number of TLS-related characteristic genes, coefficient i is the regression coefficient corresponding to the i-th TLS-related characteristic gene related to prognosis. For each GBM sample, there is a corresponding prognostic risk score.

[0070] Step S13: Based on the prognostic risk score values ​​of the GBM samples in the training data set calculated in the above steps, the GBM samples to be predicted in the GBM sample set to be predicted are divided into a high risk score group and a low risk score group according to the optimal risk score threshold.

[0071] Optionally, the above step S13 includes the following steps:

[0072] S131. Divide the samples into high and low risk score groups according to the optimal risk score threshold.

[0073] Based on the prognostic risk score value of each GBM sample in the training data set, the optimal threshold of the risk score for dividing the samples into high-risk group and low-risk group is calculated according to the statistical Youden index.

[0074] According to the optimal risk score threshold, the GBM samples in the training data set (or the GBM samples to be predicted in the GBM sample set to be predicted) are divided into a low risk score group and a high risk score group, and the prognostic risk score of each GBM sample to be predicted in the GBM sample set to be predicted is calculated. The GBM samples to be predicted whose prognostic risk score is greater than the optimal risk score threshold are divided into the high risk score group, and the GBM samples to be predicted whose score is less than the optimal risk score threshold are divided into the low risk score group.

[0075] In this embodiment, according to the optimal threshold of risk score, the GBM samples of the training data set, the GBM samples of the validation data set, and the GBM samples to be predicted in the GBM sample set to be predicted are divided into high-risk score groups and low-risk score groups, and then the subjects corresponding to each GBM sample to be predicted in the GBM sample set to be predicted are divided into high-risk score groups and low-risk score groups.

[0076] In addition, optionally, in this embodiment, the overall survival of the training data set TCGA-GBM samples and the validation data set CGGA.mRNAseq_325 samples can be analyzed to evaluate the reliability of the prognostic risk score.

[0077] In a feasible embodiment of the present invention, in the GBM samples of the training data set and the validation data set, Kaplan-Meier analysis is used to compare the difference in overall survival between the low risk score group and the high risk score group samples, and the area under the receiver operating curve (AUC) is used to evaluate the effectiveness of the risk score in predicting overall survival.

[0078] Kaplan-Meier analysis showed that in the GBM samples of the training data set and the GBM samples of the validation data set, the overall survival of the GBM samples in the high-risk score group was often significantly shorter than that of the GBM samples in the low-risk score group. In addition, the AUC values ​​of the risk score group for predicting the 1 / 3 / 5-year overall survival of the training data set samples were 0.691, 0.784, and 0.885, respectively, and the AUC values ​​for predicting the 1 / 3 / 5-year overall survival of the validation data set samples were 0.559, 0.639, and 0.779, respectively, showing good predictive efficacy ( Figure 4 , Figure 5 ).

[0079] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods.

[0080] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0081] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0082] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0083] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A prognostic risk prediction method based on tertiary lymphatic structure characteristic genes, characterized in that: The following steps are involved: Step S11, based on the TLS-related gene feature set, the transcriptional profiles of the GBM samples in the training data set are divided into different GBM molecular subtypes by a consensus clustering algorithm; Step S12, identifying differentially expressed genes in different GBM molecular subtypes, screening differentially expressed genes with prognostic value, obtaining TLS-related characteristic genes related to prognosis, and calculating the prognostic risk score of each GBM sample in the training data set; Step S13: obtaining an optimal risk score threshold according to the prognostic risk score of each GBM sample in the training data set, and dividing each GBM sample to be predicted in the GBM sample set to be predicted into a high risk score group and a low risk score group according to the optimal risk score threshold.

2. The prognostic risk prediction method based on tertiary lymphatic structure characteristic genes according to claim 1, characterized in that: The GBM samples of the training data set are obtained from the TCGA-GBM training data set samples obtained from the Cancer Genome Atlas database.

3. The prognosis risk prediction method based on tertiary lymphatic structure characteristic genes according to claim 1, characterized in that: The differentially expressed genes in different GBM molecular subtypes are obtained based on the following steps: Differentially expressed genes between different GBM molecular subtypes were screened through differential expression analysis to obtain differentially expressed genes; The differentially expressed genes with prognostic value are obtained based on the following steps: Through univariate proportional hazard Cox regression analysis, prognostic analysis was performed on all the differentially expressed genes, and multiple differentially expressed genes with prognostic value were screened; The prognosis-related TLS-related characteristic genes are obtained based on the following steps: The LASSO-Cox regression analysis algorithm was used to further screen the differentially expressed genes with prognostic value to obtain TLS-related characteristic genes related to prognosis.

4. The method for predicting prognosis risk based on tertiary lymphatic structure characteristic genes according to claim 3, characterized in that: The prognostic risk score is obtained based on the following formula: Among them, expression i is the expression level of the i-th TLS-related characteristic gene related to prognosis in the GBM sample of the training data set, n is the number of TLS-related characteristic genes, coefficient i is the regression coefficient corresponding to the i-th TLS-related characteristic gene related to prognosis.

5. The method for predicting prognosis risk based on tertiary lymphatic structure characteristic genes according to claim 4, characterized in that: The method of obtaining the optimal threshold of risk score according to the prognostic risk score of each GBM sample in the training data set is based on the following steps: Based on the prognostic risk score value of each GBM sample in the training data set, the optimal threshold of the risk score for dividing the samples into high-risk group and low-risk group is calculated according to the statistical Youden index.

6. The method for predicting prognosis risk based on tertiary lymphatic structure characteristic genes according to claim 4, characterized in that: The step of dividing each GBM sample to be predicted in the GBM sample set into a high risk score group and a low risk score group according to the optimal risk score threshold comprises the following steps: Calculating the prognostic risk score of each GBM sample to be predicted in the GBM sample set to be predicted; GBM samples to be predicted whose prognostic risk score is greater than the optimal risk score threshold are divided into the high risk score group, and GBM samples to be predicted whose score is less than the optimal risk score threshold are divided into the low risk score group.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the prediction method according to any one of claims 1 to 6 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the prediction method according to any one of claims 1 to 6 are implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the prediction method according to any one of claims 1 to 6 are implemented.

Images