Large-scale expression map correlation-based disulfide death regulation gene prediction method, system, equipment, medium and program

Abstract

The invention discloses a method, a system, equipment, a medium and a program for predicting a disulfide death regulatory gene based on large-scale expression map correlation, and belongs to the technical field of disulfide death regulatory gene prediction. An expression residual error of a transcriptome gene expression matrix formed by converting human large-scale gene expression map data is calculated by using a PEER method, and the expression residual error is used for replacing a gene expression quantity, so that a whole genome gene expression residual error matrix is obtained. Then calculating expression correlation between each gene and a known disulfide death regulation factor; and taking the correlation of the disulfide death regulatory factors as a weight factor of weighted average correlation, and calculating the weighted average correlation of the whole genome gene and the known disulfide death regulatory factors to obtain the disulfide death characteristics of the whole genome gene. And the first 1% of the gene with the highest disulfide death characteristic is used as a potential disulfide death regulating gene. The method can predict the dithiodeath tendency of various cancer types and the difference of dithiodeath characteristics in different tissues and organs.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of disulfide death regulatory gene prediction, and particularly relates to a disulfide death regulatory gene prediction method, system, device, medium and program based on large-scale expression atlas correlation. BACKGROUND

[0002] Metabolic reprogramming, as a double-edged sword, is not only an important feature of cancer development, but also a breakthrough that can be utilized in cancer treatment. Controllable cell death models play a key role in the chemotherapy of cancer. Gan research team reported a new cell death model regulated by metabolism, "disulfidptosis". This new death model is significantly different from other cell death models, such as apoptosis, necrosis, pyroptosis, autophagy, ferroptosis and cupricptosis, in the way of metabolic regulation (Liu X, Nie L, Zhang Y, Yan Y, Wang C, Colic M, Olszewski K, Horbath A, Chen X, Lei G, Mao C, Wu S, Zhuang L, Poyurovsky MV, James You M, Hart T, Billadeau D D, Chen J, Gan B. Actin cytoskeleton vulnerability to disulfide stress mediates disulfidptosis [J]. Nature Cell Biology, 2023, 25 (3): 404-414.). The current condition for inducing disulfidptosis is that by limiting glucose uptake, the pentose phosphate pathway in cells is inhibited, reduced coenzyme II (NADPH) is rapidly depleted, and high expression of SLC7A11 accelerates the accumulation of cystine in cells, thereby inducing disulfide stress, causing errors in the cross-linking of actin cytoskeleton, and ultimately leading to cell death.

[0003] Since 2017, Gan et al. found that high expression of SLC7A11 sensitized cancer cells to glucose dependence, which could lead to cell death under glucose starvation conditions (Koppula P, Zhang Y, Shi J, Li W, Gan B. The glutamate / cystine antiporter SLC7A11 / xCT enhances cancer cell dependency on glucose by exporting glutamate [J]. Journal of Biological Chemistry, 2017, 292(34): 14240-14249.). In the following years, the Gan team successively found that a series of enzymes including PGD, G6PD, TKT and TALDO1, which are involved in the glucose phosphogluconate pathway, play a role in inhibiting cell death (Liu X, Olszewski K, Zhang Y, Lim E W, Shi J, Zhang X, Zhang J, Lee H, Koppula P, Lei G, Zhuang L, You M J, Fang B, Li W, Metallo C M, Poyurovsky M V, Gan B. Cystine transporter regulation of pentose phosphate pathway dependency and disulfide stress exposes a targetable metabolic vulnerability in cancer [J]. Nature Cell Biology, 2020, 22(4): 476-486.). In addition, the activation of ATF4 and the inhibition of PRC1 can synergistically promote cancer cell death under glucose starvation conditions. In 2023, the Gan team officially named this way of cell death as "disulfide death", at the same time, they also used whole genome CRISPR / Cas9 screening to get SLC3A2, RPN1, NCKAP1, CYFIP1, WASF2, ABI2, BRK1 and RAC1, these genes promote disulfide death, and NUBPL, NDUFA11, LRPPRC, OXSM, NDUFS1 and GYS1, these regulatory factors inhibit disulfide death.

[0004] However, as a newly discovered form of regulatory cell death, our understanding of the regulatory mechanisms behind bithionol death is still very limited. Although some studies have predicted the bithionol death status of various types of tumors by known bithionol death molecular characteristics, and identified anti-tumor drugs based on the same, they only used known bithionol death genes, and lacked exploration of other regulatory factors that could cause bithionol death patterns. In addition, some studies have constructed risk models for related diseases by using omics data of specific cancers in the TCGA database, and have found some prognostic characteristics, but the amount of data used in these studies may make it difficult to discover more general bithionol death regulatory factors, and the overall picture of bithionol death cannot be seen. In summary, there is currently a lack of model methods that can predict bithionol death regulatory factors from a whole genome perspective, which limits the precise treatment of cancer. SUMMARY

[0005] In order to overcome the above-mentioned shortcomings of the prior art, the purpose of the present application is to provide a bithionol death regulatory gene prediction method, system, device, medium and program based on large-scale expression profile correlation, to make new predictions of bithionol death regulatory factors.

[0006] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions to achieve it:

[0007] The first aspect of the present application discloses a bithionol death regulatory gene prediction method based on large-scale expression profile correlation, comprising:

[0008] Collecting human large-scale gene expression profile data and known bithionol death regulatory factor data;

[0009] Converting the obtained human large-scale gene expression profile data into a transcriptome gene expression matrix; calculating the expression residual of the transcriptome gene expression matrix using the PEER method, using the expression residual instead of the gene expression amount, and obtaining a whole genome gene expression residual matrix;

[0010] Using the obtained whole genome gene expression residual matrix to calculate the expression correlation between each gene and the collected known bithionol death regulatory factor;

[0011] Taking the correlation of the bithionol death regulatory factor as a weight factor of the weighted average correlation, obtaining the bithionol death characteristics of the whole genome gene by calculating the weighted average correlation of the whole genome gene and the known bithionol death regulatory factor, and taking the top 1% genes with the highest bithionol death characteristics of the whole genome gene as potential regulatory bithionol death genes.

[0012] Preferably, the human large-scale gene expression profile data is transcriptome data.

[0013] Preferably, the known bithionol death regulatory factor data includes regulatory factors promoting bithionol death and regulatory factors inhibiting bithionol death.

[0014] Further preferably, the known bithionol death regulatory factor data includes SLC7A11, SLC3A2, RPN1, NCKAP1, CYFIP1, WASF2, ABI2, BRK1, RAC1, ATF4, NUBPL, NUDUFA11, LRPPRC, OXSM, NDUFS1, GYS1, G6PD, PGD, TALDO1, TKT, PRC1, SLC2A1, SLC2A2, SLC2A3 and SLC2A4.

[0015] Preferably, after obtaining the expression residual of the transcriptome gene expression matrix, the dataset with a data amount less than 80 samples is removed, and the expression residual is used to replace the gene expression amount to obtain a whole genome gene expression residual matrix.

[0016] Preferably, the expression correlation analysis is performed by using a Pearson correlation coefficient or a Spearman correlation coefficient.

[0017] Preferably, in the process of calculating the expression correlation between each gene and the collected known bithionol death regulatory factor, a Bonferroni correction is used to eliminate the false discovery rate.

[0018] The second aspect of the present application discloses a bithionol death feature evaluation method, comprising:

[0019] The weighted average of the expression levels of each gene of the target disease sample, the target tissue sample or the target organ sample and the bithionol death feature value obtained by the above bithionol death regulatory gene prediction method is taken to obtain the weighted bithionol death feature value of the target disease sample, the target tissue sample or the target organ sample and the normal tissue;

[0020] The weighted bithionol death feature values of the target disease sample, the target tissue sample or the target organ sample and the normal tissue are compared, and the cancer types with the number of tumor group and normal group samples with non-zero gene expression and corresponding bithionol death features greater than 3 are screened to obtain the bithionol death feature evaluation result of the target disease sample, the target tissue sample or the target organ sample.

[0021] The third aspect of the present application discloses a bithionol death regulatory gene prediction system based on large-scale expression atlas correlation, comprising:

[0022] The acquisition module is used to acquire human large-scale gene expression atlas data and known bithionol death regulatory factor data.

[0023] The processing module is used for converting human large-scale gene expression profile data into a transcriptome gene expression matrix, and calculating expression residuals of the transcriptome gene expression matrix by using a PEER method to obtain a whole genome gene expression residual matrix;

[0024] The analysis module is used for calculating expression correlations between each gene and known double sulfur death regulatory factors by using the whole genome gene expression residual matrix.

[0025] The prediction module is used for taking the correlation of the double sulfur death regulatory factor as a weight factor of a weighted average correlation, obtaining a double sulfur death feature of the whole genome gene by calculating the weighted average correlation between the whole genome gene and the known double sulfur death regulatory factor, and taking the top 1% genes with the highest double sulfur death feature of the whole genome gene as potential double sulfur death regulatory genes.

[0026] In a fourth aspect, the present application discloses a computer device, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the double sulfur death regulatory gene prediction method based on large-scale expression profile correlation.

[0027] In a fifth aspect, the present application discloses a computer readable storage medium, which stores a computer program, and the computer program is executable on a processor to implement the steps of the double sulfur death regulatory gene prediction method based on large-scale expression profile correlation.

[0028] Compared with the prior art, the present application has the following beneficial effects:

[0029] A method for predicting disulfide death regulatory genes based on large-scale expression map correlations: 1) Collect large-scale human gene expression map data and known disulfide death regulatory factor data. Combining known disulfide death regulatory factors with large-scale human gene expression map data has two advantages: First, the large sample size of large-scale human gene expression map data can significantly reduce the impact of random fluctuations and improve the reliability and robustness of correlation estimation, especially for gene pairs with low expression or small expression differences, where weak correlation signals are easier to identify; second, the data comes from different experiments and tissue types, making the calculated correlations (co-expression regulation) more widely applicable (generalization ability); 2) Convert the obtained large-scale human gene expression map data into a transcriptome gene expression matrix; use the PEER method to calculate the expression residuals of the transcriptome gene expression matrix. The process involves using expression residuals instead of gene expression levels to obtain a genome-wide gene expression residual matrix. In this step, the gene expression heterogeneity handling method (PEER method for residual estimation) reduces the heterogeneity issues inherent in large-scale human transcriptome data from different research purposes and experimental designs, improves the sensitivity and interpretability of gene associations in large-scale gene expression data, effectively enhances the accuracy of disulfidptosis feature models, and more realistically reflects the differences and associations between gene expressions. 3) The obtained genome-wide gene expression residual matrix is ​​used to calculate the expression correlation between each gene and collected known disulfidptosis regulatory factors. The aim is to systematically mine potential new regulatory genes that are expression-correlated with known disulfidptosis regulatory factors based on the genome-wide residual expression matrix. By comparing the synergistic expression change patterns of each gene and key disulfidptosis factors, candidate genes that may jointly participate in or regulate the disulfidptosis pathway can be identified from high-dimensional gene expression data. Using residual matrices instead of raw expression data effectively removes potential batch effects, tissue-specific differences, or other known / unknown covariates between samples. This allows the calculated correlations to better reflect the intrinsic co-expression relationships between genes, rather than epigenetic correlations interfered with by covariates. This method improves the specificity and biological interpretability of association analysis while eliminating background noise. 4) The correlation of disulfide death regulators is used as a weighting factor in the weighted average correlation. By calculating the weighted average correlation between genome-wide genes and known disulfide death regulators, the disulfide death characteristics of genome-wide genes are obtained. The aim is to construct a more systematic and robust disulfide death characteristic scoring method by comprehensively considering the expression correlation between genome-wide genes and known disulfide death regulators and introducing a weighting mechanism based on the correlation of the regulators themselves, in order to accurately identify potential novel disulfide death regulator genes.Specifically, by introducing the correlation between regulatory factors as a weighting coefficient, the calculated weighted average correlation not only retains the individual association information between genes and each regulatory factor, but also fully considers the interaction and redundancy between regulatory factors in the biological network, thus constructing a more representative integrated index in biology. Compared with simple averaging or maximum correlation strategies, the weighted average correlation better reflects the overall "affinity" of the target gene in the disulfide death regulatory network. The "disulfide death feature score" obtained based on this weighted correlation can be regarded as a probability score of each gene participating in or regulating the disulfide death pathway. 5) The top 1% of genes with the highest disulfide death features in the whole genome are selected as potential disulfide death regulatory genes; this can effectively screen candidate regulatory factors closely related to disulfide death across the whole genome, laying the foundation for subsequent functional verification, mechanism research, and target discovery. Furthermore, this method for predicting disulfide death regulatory genes can quantitatively predict the degree of disulfide death in specific cellular states; it can also predict novel disulfide death regulators, discover disulfide death regulators and induction mechanisms, and screen candidate targeted drugs for cancer treatment; and it can quantitatively evaluate the degree of disulfide death in specific cellular states, providing broad application prospects for future precision cancer treatment. The discovery of consistent gene expression patterns among existing disulfide death regulators in large samples provides further support for the future discovery of new disulfide death-mediated processes.

[0030] Furthermore, when performing multiple comparison adjustments, using Bonferroni correction to eliminate the false discovery rate can solve the problem of false positives when estimating the correlation of a large number of samples. Attached Figure Description

[0031] Figure 1 Gene expression correlation matrix of known disulfide death regulators;

[0032] Figure 2 This section describes the construction of the disulfidptosis signature (DS). Here, a represents the Pearson and Spearman correlation coefficients of gene expression; b represents the frequency distribution of Pearson correlation coefficients between 25 known disulfidptosis regulators and other genes; c represents the regulatory factor weights used to construct the DS; d represents the distribution of standardized Pearson correlation coefficients between disulfidptosis regulators and other genes, sorted from highest to lowest by average correlation value; and e represents the frequency distribution of DS scores.

[0033] Figure 3Differences in weighted disulfidptosis signature (WDS) between normal and tumor groups in different cancer types;

[0034] Figure 4 Ranking of WDS in different cancer types;

[0035] Figure 5 Comparison of survival curves between high DS and low DS groups in patients with different cancer types;

[0036] Figure 6 To sort WDS among different tissues or organs;

[0037] Figure 7 To investigate sex differences in DS in different tissues or organs;

[0038] Figure 8 The first image shows the characteristic importance network of the top 1% of DS genes and the effects of mimic gene overexpression and CRISPR-Cas9 knockout perturbation on DS genes; image a shows the characteristic importance network of the top 1% of DS genes, where pink indicates positive regulators and blue indicates negative regulators. Other genes are shown with their DS size in a gradient from blue-green to yellow. The thickness of the lines connecting nodes indicates their interaction score in the STRING database; image b shows the pathway enrichment analysis of the top 1% of DS genes. Red arrows indicate cytoskeleton-related pathways, blue arrows indicate energy metabolism and disease-related pathways, and yellow arrows indicate oxidative stress-related pathways; image c shows the pathway enrichment network of the top 1% of DS genes, where node size indicates the number of genes involved in the pathway, and node color indicates the significance of the pathway (p-value); image d shows the DS gene heatmap of the top 500 and bottom 500 samples ranked by MWDS after mimic gene overexpression perturbation. The horizontal axis represents single-gene overexpression samples, sorted from high to low MWDS. Light pink indicates MWDS>0 (positive), and light green indicates MWDS<0 (negative). The vertical axis represents DS, sorted from high to low. Red indicates high DS, and blue indicates low DS. e is a heatmap of DS scores for the top 500 and bottom 500 samples after simulated gene CRISPR knockout perturbation. The horizontal axis represents single-gene knockout samples, sorted from low to high MWDS. f is a heatmap of genes with the most significant impact on MWDS in simulated gene overexpression and CRISPR knockout perturbation. g is a ridge plot of simulated gene overexpression perturbation, with black dots representing the average. h is a ridge plot of simulated gene CRISPR knockout perturbation, with black dots representing the average. i is the six genes with the most significant impact on MWDS in simulated overexpression perturbation. j is the six genes with the most significant impact on MWDS in simulated CRISPR knockout perturbation. Detailed Implementation

[0039] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0040] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0041] The present invention provides a method for predicting disulfide death regulatory genes based on large-scale expression map correlation, comprising the following steps:

[0042] S1. Data Collection

[0043] Large-scale human gene expression mapping data and known disulfide death regulator data were collected. The large-scale human gene expression mapping data came from different experimental and tissue types, and the known disulfide death regulator data included genes that inhibit disulfide death and genes that promote disulfide death.

[0044] S2, Data Processing

[0045] The large-scale human gene expression map data obtained in S1 was converted into a transcriptome gene expression matrix. The expression residuals of the transcriptome gene expression matrix were calculated using the PEER (Probabilistic Estimation of Expression Residuals) method to remove confounding factors. Data sets with fewer than 80 samples were removed, and the expression residuals were used instead of gene expression levels to obtain a genome-wide gene expression residual matrix.

[0046] S3, Expression Correlation Analysis

[0047] The expression correlation between each gene and the known disulfide death regulator data collected in S1 was calculated using the genome-wide gene expression residual matrix obtained in S2.

[0048] S4. Obtain the disulfidptosis signature (DS) of the entire genome.

[0049] The correlation of disulfide death regulators was used as the weighting factor for the weighted average correlation. The disulfide death characteristics of whole-genome genes were obtained by calculating the weighted average correlation between genes in the whole-genome gene expression residual matrix and known disulfide death regulator data. The top 1% of genes with the highest disulfide death characteristics in whole-genome genes were selected as potential genes regulating disulfide death.

[0050] The method for assessing disulfide mortality characteristics provided by this invention is characterized by comprising:

[0051] 1. Take the weighted average of the expression levels of each gene in the target disease sample, target tissue sample, or target organ sample and the disulfide mortality characteristic value of that gene to obtain the weighted disulfide mortality characteristic value of the target disease sample, target tissue sample, or target organ sample and normal tissue.

[0052] 2. Compare the weighted disulfide mortality characteristics of target disease samples, target tissue samples, or target organ samples with those of normal tissues, and screen out cancer types with more than 3 tumor groups and normal groups that retain non-zero gene expression and corresponding disulfide mortality characteristics to obtain the disulfide mortality characteristic assessment results of target disease samples, target tissue samples, or target organ samples.

[0053] The following detailed description of the invention, combined with specific methods for predicting and evaluating disulfide death regulatory genes based on large-scale expression map correlations, is intended to explain rather than limit the invention.

[0054] Example 1

[0055] A method for predicting disulfide death regulatory genes based on large-scale expression map correlations includes the following steps:

[0056] S1. Data Collection

[0057] S1.1, Collect large-scale human transcriptome data

[0058] Human large-scale expression atlas data were collected and compiled from the GeneBridge website. The data covers 272,445 different tissues and organs from 1,454 datasets collected from the GEO database, as well as human gene expression levels including samples from different disease types.

[0059] S1.2 Obtain known disulfide death regulators

[0060] The following 25 genes (details shown in Table 1) were used as known disulfide death regulators; based on their different roles in the disulfide death regulation process, they were divided into promoting and inhibiting components, as detailed below:

[0061] The ten regulatory factors that promote disulfide death are: SLC7A11, SLC3A2, RPN1, NCKAP1, CYFIP1, WASF2, ABI2, BRK1, RAC1 and ATF4.

[0062] The regulatory factors that inhibit disulfide death (15 in total) are: NUBPL, NUDUFA11, LRPPRC, OXSM, NDUFS1, GYS1, G6PD, PGD, TALDO1, TKT, PRC1, SLC2A1, SLC2A2, SLC2A3 and SLC2A4.

[0063] Table 1 Information on 25 known disulfide death regulators

[0064]

[0065]

[0066] S2, Data Processing

[0067] S2.1 Constructing the transcriptome gene expression matrix

[0068] Gene expression data from 272,445 samples from 1,454 studies were combined into a single gene expression matrix to obtain the transcriptome gene expression matrix.

[0069] S2.2 Estimation of gene expression residuals

[0070] The expression residuals of the transcriptome gene expression matrix were calculated using the PEER method. Data sets with fewer than 80 samples were removed, and the expression residuals were used to replace gene expression levels to obtain the genome-wide gene expression residual matrix.

[0071] Among the methods described above, the advantage of using residuals instead of expression levels is that it removes the influence of systematic confounding factors such as sample source, technical noise, and experimental batch, and can more realistically reflect the differences and correlations between gene expressions.

[0072] S3, Expression Correlation Analysis

[0073] S3.1 Calculate the expression correlation

[0074] Based on the Pearson correlation coefficient calculation formula (Formula (1)), the expression correlation coefficients between 25 known disulfide death regulators and the expression correlation coefficients between each gene in the whole genome and the 25 known disulfide death regulators were calculated using the genome-wide gene expression residual matrix obtained in S2.2.

[0075]

[0076] Where, x i ,y i For the i-th observation; Let x and y be the mean; n be the sample size; r be the mean of x and y. xy ∈[-1,1]: The closer to ±1, the stronger the linear relationship.

[0077] Alternatively, for x in the whole-genome gene expression residual matrix obtained in S2.2 i y i Sort them separately to obtain the rank R(x) i ),R(y i Then, according to the Spearman correlation coefficient calculation formula (Formula (2)), the expression correlation coefficients between 25 known disulfide death regulators and the expression correlation coefficients between each gene in the whole genome and 25 known disulfide death regulators are calculated.

[0078]

[0079] Where, d i =R(x) i )-R(y i ): The rank difference of the i-th sample between the two variables; n: The number of samples; ρ∈[-1,1]: The closer to ±1, the stronger the monotonic relationship.

[0080] The results of using the Pearson correlation coefficient and the Spearman correlation coefficient, respectively, are as follows: Figure 2 As shown, the correlations between the genes calculated by the two methods are very close (R0). 2 =0.98, p<1*10^-308)( Figure 2 (a) Most genes showed low correlation, with an average correlation of 0.019. Figure 2 (b)

[0081] S3.2 Eliminating False Discovery Rate

[0082] The calculation results obtained in S3.1 were adjusted for multiple comparisons and then corrected using Bonferroni. Assuming that the number of samples for comparing gene expression correlations is samples, the significance level for each test is 0.05 / samples, and the expression correlations between each gene and 25 known disulfide death regulators were obtained.

[0083] The results of the expression correlation analysis are as follows Figure 1 As shown, the expression correlations among the disulfide death regulators G6PD, PGD, TALDO1, and TKT were the highest, followed by the correlation between SLC7A11 and SLC3A2, which was also higher than that of other genes. The correlations among known disulfide death regulators were significantly higher than those among other genes in the genome (p = 1.28 * 10^9). -165 ).

[0084] S4. Obtain disulfide death features from the entire genome.

[0085] S4.1 Calculate the weighted average correlation

[0086] The weighted average correlation between genes in the genome-wide gene expression residual matrix and 25 known disulfide death regulators is calculated according to formula (3), and the correlation of genes with inhibition is negative.

[0087]

[0088] Among them, Similarity j This represents the expression correlation value between other genes in the genome and the j-th known disulfide death regulator. Adjusted Similarity j The criteria for determining the corrected correlation value are shown in formula (4):

[0089]

[0090] w j The weight of the j-th known disulfide mortality regulator is calculated using the formula shown in formula (5):

[0091]

[0092] Where m is the total number of genes with known disulfide death regulators. For the i-th gene, the disulfide death characteristic of the gene is obtained by normalization using the z-score of the whole genome, as shown in formula (6):

[0093]

[0094] Where μ is the Weighted Average Correlation of the i-th gene in the whole genome. i The mean, σ is the weighted average correlation of the i-th gene in the whole genome. i The standard deviation.

[0095] S4.2 Obtain the characteristic distribution of disulfide death: such as Figure 2 As shown in Figure c, the correlation of disulfide death regulators obtained in S3 was used as the weighting factor for the weighted average correlation to compare and rank the expression correlations of disulfide death regulators with other genes. Figure 2 (d) , to obtain the distribution of disulfide death features of the whole genome ( Figure 2 (e).

[0096] S4.3. The top 1% of genes (n=304) with the highest disulfide death characteristics in the disulfide death characteristic distribution of S4.2 were selected as potential regulatory genes for disulfide death. The results are shown in Table 2.

[0097] Table 2 Potential genes regulating disulfide death

[0098]

[0099]

[0100]

[0101]

[0102]

[0103] Example 2

[0104] Example 2 provided by the present invention is an example of evaluating disulfide mortality characteristics based on TCGA cancer samples.

[0105] To assess the characteristic variations of disulfide death across different cancer types, a comprehensive analysis of disulfide death characteristics in tumor and normal tissues across multiple cancer types was conducted. The degree of disulfide death in tumor and normal tissues was represented using weighted disulfide death signatures (WDS). Cancer types lacking normal tissue or with insufficient tissue samples were not discussed.

[0106] 1. We obtained 11,274 complete RNA-seq and clinical data for 33 cancer types from the TCGA database website (https: / / portal.gdc.cancer.gov / ). The weighted average density (WDS) of tumor and normal tissues for each cancer was calculated by weighting the expression levels (TPM) of each gene and the DS value of that gene in each sample. A weighted disulfide mortality feature was calculated for each sample. When comparing the weighted disulfide mortality features of the tumor and normal groups for each cancer type, we selected and retained cancer types with more than 3 samples in each group and one group with corresponding disulfide mortality features, totaling 21 types.

[0107] The results showed that in 15 cancer types, the WDS in tumor tissue was significantly lower than that in normal tissue, suggesting that the disulfide death pathway may be suppressed in these tumors. Figure 3 Cervical squamous cell carcinoma (CESC) showed the greatest reduction in WDS (Mean Diff: -0.67, P-value: 3.34e-15), indicating a significantly enhanced inhibitory effect of disulfide death in CESC tumor tissue. This inhibitory phenomenon was also prominent in endometrial carcinoma (UCEC) (Mean Diff: -0.56, P-value: 6.83e-54) and lung squamous cell carcinoma (LUSC) (Mean Diff: -0.52, P-value: 5.27e-69), suggesting that disulfide death inhibition may play a key role in tumor progression in these tumor types. In contrast, cholangiocarcinoma (CHOL) (Mean Diff: 0.37, P-value: 4.44e-08), pheochromocytoma and paraganglioma (PCPG) (Mean Diff: 0.16, P-value: 2.35e-02), and clear cell renal cell carcinoma (KIRC) (Mean Diff: 0.13, P-value: 1.71e-08) showed elevated WDS in tumor tissues. Interestingly, even though these tumor tissues exhibited the opposite trend of disulfide death, their overall average WDS remained less than zero, indicating that tumor tissues do not tend to promote disulfide death. This phenomenon suggests that disulfide death in different cancer types may have specific regulatory mechanisms in the tumor microenvironment.

[0108] 2. By sorting the WDS of 33 cancer types from smallest to largest, we can visually compare the tendency of these cancer types to die from disulfides.

[0109] WDS comparison among 33 cancer types Figure 4As shown, among all cancer types, hepatocellular carcinoma (LIHC), uveal melanoma (UVM), and rectal adenocarcinoma (READ) had significantly lower WDS than other cancer types, suggesting that tumor cells in these cancer types may rely on the inhibition of disulfide death to promote survival. Furthermore, sarcoma (SARC), clear cell renal cell carcinoma (KIRC), and mesothelioma (MESO) had relatively higher WDS, suggesting that the disulfide death pathway may play a more active role in tumor metabolic regulation and growth adaptation in these cancer types. The differences in WDS among different cancer types reveal the crucial role of disulfide death in tumor metabolic regulation. Some cancers, such as LIHC and CESC, may enhance tumor cell survival by inhibiting the disulfide death pathway, while other cancers, such as KIRC, may maintain tumor cell metabolic homeostasis by activating the disulfide death pathway.

[0110] 3. Compare the survival probability curves of patients with high disulfide mortality characteristics (High DS) and low disulfide mortality characteristics (Low DS) in various cancers, covering multiple tumor types. High DS and Low DS refer to the top 1% and bottom 1% of genes, respectively.

[0111] Each chart shows the change in survival probability over time for different cancer patient groups, and assesses the impact of different DS levels on patient survival using Kaplan-Meier curves. The red curve represents patients with high DS, and the blue curve represents patients with low DS. Figure 5 In most cancers, patients with high disulfide death (DS) exhibit better survival rates, which may suggest that disulfide death mechanisms in these cancers are related to cell protection or anti-cancer activity, such as KIRC, BRCA, and LUAD. In some cancers, the difference in survival rates between patients with high and low DS is small, indicating that disulfide death characteristics have a weak or no significant impact on the prognosis of these cancer patients, such as CESC and STAD.

[0112] Example 3

[0113] Example 3 provided by the present invention is an example of evaluating disulfide mortality characteristics based on different tissues and organs.

[0114] To further compare the regulation of the disulfide death pathway in different normal tissues and organs, we analyzed the WDS in different normal tissues and organs.

[0115] Gene expression data from different tissues and organs were obtained from the GTEx database (https: / / gtexportal.org / home / ). The WDS (Wide Distributed Data Set) of gene expression levels was used to represent the disulfide mortality tendency of a specific tissue sample. Tissues with more than three samples were retained for comparison of disulfide mortality characteristics. Figure 6The distribution of water dispersive molecule (WDS) across different normal tissues and organs was presented. The results showed significant differences in WDS among different normal tissues and organs, suggesting that the regulation of disulfide death in these tissues is highly tissue-specific. Specifically, organs such as the kidney, liver, and gallbladder had higher WDS values, suggesting that these tissues may maintain cellular metabolic homeostasis and redox balance by activating the disulfide death pathway under normal physiological conditions. For example, the kidney, as a metabolically active organ, may need to maintain intracellular protein folding and stability through disulfide bond formation and regulation. Organs such as the spleen and pancreas had lower WDS values, suggesting that these organs may rely less on the disulfide death pathway for cell death regulation under normal conditions, and rely more on other types of cell death pathways. Of particular note was the relatively low WDS value in lung tissue, consistent with the WDS pattern in lung cancer, indicating that lung tissue may rely less on the disulfide death pathway under both normal and pathological conditions. This may be related to the unique metabolic needs of lung tissue and the oxidative stress response in its microenvironment.

[0116] By performing WDS analysis on gene expression levels of different sexes in various human tissues and organs from the GTEx database, the differences in WDS distribution between sexes in each tissue were obtained, and the statistical differences were assessed using a significance test (P-value). Figure 7 From the perspective of p-value significance, the sex differences in adipose tissue (Diff = 1.05, P = 2.73e-06), brain (Diff = -0.61, P = 1.42e-06), heart (Diff = 0.8, P = 2.58e-04), muscle (Diff = -0.59, P = 1.41e-03), nerve (Diff = 0.52, P = 8.50e-03), salivary glands (Diff = 1.87, P = 4.69e-02), small intestine (Diff = 1.29, P = 3.96e-02), and spleen (Diff = -1.33, P = 4.53e-03) were relatively significant, indicating that sex had a greater influence on disulfide mortality characteristics in these tissues. Among them, salivary glands and spleen showed a greater degree of sex difference in disulfide mortality (difference values ​​were 1.87 and -1.33, respectively). In muscle and nerve tissues, WDS was significantly higher in males than in females, while the opposite was true for other tissues. In most digestive system organs (such as the esophagus, stomach, kidneys, and pancreas) and endocrine organs (such as the thyroid gland and pituitary gland), WDS showed little difference between sexes, with large p-values, indicating that the disulfide mortality characteristics in these tissues did not differ significantly between sexes.

[0117] By comparing the changing trends of weighted disulfide mortality characteristics (WDS) in various tissues and organs across different age groups (20–80 years, grouped at ten-year intervals), we can further understand the potential impact of age on disulfide mortality characteristics. Figure 8In most tissues and organs, such as adipose tissue, brain, esophagus, heart, and liver, WDS values ​​remained relatively stable across age groups, showing no significant age-related changes. This indicates that these tissues have low sensitivity to age-related factors, and disulfide mortality characteristics were not significantly affected by age effects at different age stages of adulthood. However, elevated WDS values ​​were observed in tissues such as the lungs, prostate, salivary glands, and small intestine at older ages (60–79 years), suggesting an age-related correlation of disulfide mortality characteristics in these tissues. For the cervix and fallopian tubes, WDS values ​​fluctuated significantly between the 20–29 and 30–39 years age groups, but subsequently stabilized. This may reflect the influence of hormonal changes related to fertility or other specific physiological factors on disulfide mortality characteristics in younger individuals, while these characteristics gradually stabilize in subsequent age groups. Vagina and Fallopian Tube: These reproductive tissues show a certain degree of increase in WDS at an older age (50-69 years old), which may be related to menopause or physiological changes in old age, suggesting that the disulfide death characteristics of these tissues may be affected by changes in sex hormones in old age.

[0118] In summary, the differences in WDS across different normal tissues reveal the tissue-specific functions of disulfide death in different organs. These differences not only reflect variations in metabolic demands and oxidative stress responses across organs but also suggest that disulfide death may play a more significant role in homeostatic regulation in certain tissues. These findings provide important insights for further research into the regulatory mechanisms of disulfide death under different physiological and pathological conditions.

[0119] Example 4

[0120] Example 4 provided by this invention is an example of evaluating disulfide death characteristics based on LINCS L1000 simulated gene overexpression and knockout perturbation data.

[0121] Simulated gene overexpression and CRISPR knockout perturbation data were obtained from RNA-seq predicted gene expression characteristics data based on Cycle-Consistent Generative Adversarial Network (CycleGAN) from the NIH LINCS program (https: / / lincsproject.org / ). The top 1% and bottom 1% of genes with the highest and lowest disulfide death characteristics were used to calculate the changes in disulfide death characteristics for each sample under overexpression (34,164 samples) and CRISPR knockout (140,946 samples) perturbation conditions. Similarly, the weighted average disulfide death characteristics of these genes were used to represent the average disulfide death level of the perturbation sample. When comparing high-DS and low-DS gene perturbations, the z-score of the average weighted disulfide death characteristics was calculated to represent the standardized average weighted DS.

[0122] Gene feature importance networks were constructed using the top 1% of genes with disulfide death features, and the networks were visualized using Cytoscape. Figure 8 (a) In the network, pink nodes represent positively regulating genes, and blue nodes represent negatively regulating genes. The colors of other genes indicate the magnitude of their disulfide death eigenvalues. The size of the nodes in the network represents the number of interactions between genes; larger nodes indicate a higher degree of interaction between these genes and other genes. The feature importance network reflects that some genes with high disulfide death eigenvalues ​​occupy important node positions in the network, such as NRF2, SIRT1, CTNNB1, and ITGB1. The genes at these important node positions have very high connectivity with disulfide death regulators and other genes, indicating that they may play a potentially important role in regulating disulfide death.

[0123] Pathway enrichment analysis was performed using KEGG on genes exhibiting high disulfide death (top 1%). Figure 8(b) This study further explored the biological pathways involved in these genes. The pathways were enriched and ranked according to GeneRatio. The pathways with the highest GeneRatio scores were significantly enriched in cytoskeleton-related pathways, such as Focal adhesion, Regulation of actin cytoskeleton, and Adherens junction. Additionally, several pathways related to glycolysis were significantly enriched, such as the Pentose phosphate pathway and Choline metabolism in cancer. Other significantly enriched pathways may be involved in cellular oxidative stress processes, such as the PI3K-Akt signaling pathway, TGF-betasignaling pathway, TNF signaling pathway, and Phospholipase D signaling pathway. The pathway network presented here illustrates the associations and interactions among these significantly enriched pathways. Figure 8 (c) The node size in the network represents the number of genes involved in that pathway, and the connections between different pathways represent gene overlap and interactions. Some pathways are located at key nodes in the network, and the connections between these pathways are also very close, such as: Focaladhesion, Regulation of actin cytoskeleton, PI3K-Akt signaling pathway, and Cholinemetabolism in cancer.

[0124] By overexpressing and CRISPR-knocking all genes with positive and negative simulated disulfide death characteristics, the average weighted disulfide death characteristic (WDS) of the genome in the sample was calculated after overexpression and CRISPR knockout. High DS represents the top 1% of genes with the highest disulfide death characteristics, and low DS represents the bottom 1% of genes with the lowest disulfide death characteristics. We ranked the genes using the weighted DS values ​​of the first and last 500 samples after simulated gene overexpression and CRISPR knockout. The results show that overexpressing genes with positive disulfide death characteristics increases the average weighted disulfide death characteristic. Conversely, overexpressing negatively regulated genes decreases the average disulfide death characteristic. Furthermore, the WDS of high DS genes is higher than that of low DS genes. Figure 8 d). CRISPR knockout yielded the opposite results: knocking out genes with negative disulfide death trait decreased the mean weighted disulfide death trait, while knocking out positively regulating genes increased the mean disulfide death trait. Furthermore, the WDS of high-DS genes was lower than that of low-DS genes. Figure 8(e). This suggests that changes in disulfide death characteristics largely depend on the expression regulation of specific genes. For gene samples with both knockout and overexpression data, the corresponding genes are sorted from high to low according to the average weighted disulfide death characteristics of the samples after overexpression minus CRISPR. It can be seen that from top to bottom, these are the genes that potentially regulate disulfide death from strongest to weakest. Figure 8 (f)

[0125] In simulated gene overexpression perturbations, genes with high disulfide death trait showed a significant difference in mean WDS compared to genes with low disulfide death trait. Figure 8 Mean Diff: 1.14, P-value: 3.30e-08. This indicates that overexpression of high disulfide death genes significantly promotes disulfide death. Conversely, in simulated CRISPR gene knockout perturbation, the mean wDS of high disulfide death genes was significantly lower than that of low disulfide death genes. Figure 8 Mean Diff: -1.54, P-value: <2e-16). This further demonstrates that knocking out high-DS genes can significantly inhibit the disulfide death process in cells. Simulated overexpression perturbation revealed six genes that most significantly affected WDS; overexpression of these genes could significantly increase WDS values, suggesting their potential promoting role in the regulation of disulfide death. Figure 8 Similarly, simulated CRISPR knockout perturbations revealed the six genes with the greatest impact on WDS; knocking out these genes significantly reduced WDS values, indicating their key repressive role in the disulfide death pathway. Figure 8 (j).

[0126] Example 5

[0127] Example 5 of the present invention is an example of the disulfide death regulatory gene prediction system based on large-scale expression map correlation provided by the present invention. The system includes an acquisition module, a processing module, an analysis module and a prediction module.

[0128] The acquisition module is used to acquire large-scale human gene expression map data and known disulfide death regulator data;

[0129] The processing module is used to convert large-scale human gene expression map data into transcriptome gene expression matrix, and to calculate the expression residuals of transcriptome gene expression matrix using the PEER method to obtain whole-genome gene expression residual matrix.

[0130] The analysis module is used to calculate the expression correlation between each gene and known disulfide death regulators using the genome-wide gene expression residual matrix;

[0131] The prediction module is used to use the correlation of disulfide death regulators as the weighting factor of the weighted average correlation. By calculating the weighted average correlation between whole-genome genes and known disulfide death regulators, the disulfide death characteristics of whole-genome genes are obtained. The top 1% of genes with the highest disulfide death characteristics of whole-genome genes are selected as potential genes regulating disulfide death.

[0132] It is understood that the disulfide death regulation gene prediction system based on large-scale expression map correlation provided by this invention corresponds to the disulfide death regulation gene prediction method based on large-scale expression map correlation provided in the foregoing embodiments. The relevant technical features of the disulfide death regulation gene prediction system based on large-scale expression map correlation can refer to the relevant technical features of the disulfide death regulation gene prediction method based on large-scale expression map correlation, specifically including the following steps:

[0133] Collect large-scale human gene expression mapping data and data on known disulfide death regulators;

[0134] The obtained large-scale human gene expression map data were converted into transcriptome gene expression matrix; the expression residuals of the transcriptome gene expression matrix were calculated using the PEER method to obtain the genome-wide gene expression residual matrix.

[0135] The expression correlation between each gene and the collected known disulfide death regulators was calculated using the obtained whole-genome gene expression residual matrix;

[0136] The correlation of disulfide death regulators was used as the weighting factor for the weighted average correlation. By calculating the weighted average correlation between whole-genome genes and known disulfide death regulators, the disulfide death characteristics of whole-genome genes were obtained. The top 1% of genes with the highest disulfide death characteristics of whole-genome genes were selected as potential genes regulating disulfide death.

[0137] Example 6

[0138] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor performs the steps of the method for predicting disulfide death regulatory genes based on large-scale expression map correlation.

[0139] Collect large-scale human gene expression mapping data and data on known disulfide death regulators;

[0140] The obtained large-scale human gene expression map data were converted into transcriptome gene expression matrix; the expression residuals of the transcriptome gene expression matrix were calculated using the PEER method to obtain the genome-wide gene expression residual matrix.

[0141] The expression correlation between each gene and the collected known disulfide death regulators was calculated using the obtained whole-genome gene expression residual matrix;

[0142] The correlation of disulfide death regulators was used as the weighting factor for the weighted average correlation. By calculating the weighted average correlation between whole-genome genes and known disulfide death regulators, the disulfide death characteristics of whole-genome genes were obtained. The top 1% of genes with the highest disulfide death characteristics of whole-genome genes were selected as potential genes regulating disulfide death.

[0143] Example 7

[0144] This embodiment provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for predicting disulfide death regulatory genes based on large-scale expression map correlation.

[0145] The method for predicting disulfide death regulatory genes based on large-scale expression map correlation includes the following steps:

[0146] Collect large-scale human gene expression mapping data and data on known disulfide death regulators;

[0147] The obtained large-scale human gene expression map data were converted into transcriptome gene expression matrix; the expression residuals of the transcriptome gene expression matrix were calculated using the PEER method to obtain the genome-wide gene expression residual matrix.

[0148] The expression correlation between each gene and the collected known disulfide death regulators was calculated using the obtained whole-genome gene expression residual matrix;

[0149] The correlation of disulfide death regulators was used as the weighting factor for the weighted average correlation. By calculating the weighted average correlation between whole-genome genes and known disulfide death regulators, the disulfide death characteristics of whole-genome genes were obtained. The top 1% of genes with the highest disulfide death characteristics of whole-genome genes were selected as potential genes regulating disulfide death.

[0150] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0151] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0152] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0153] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting disulfide death regulatory genes based on large-scale expression map correlation, characterized in that, include: Collect large-scale human gene expression mapping data and data on known disulfide death regulators; The obtained large-scale human gene expression map data was converted into a transcriptome gene expression matrix; The expression residuals of the transcriptome gene expression matrix were calculated using the PEER method. The expression residuals were then used to replace gene expression levels to obtain the genome-wide gene expression residual matrix. The expression correlation between each gene and the collected known disulfide death regulators was calculated using the obtained whole-genome gene expression residual matrix; The correlation of disulfide death regulators was used as the weighting factor for the weighted average correlation. By calculating the weighted average correlation between whole-genome genes and known disulfide death regulators, the disulfide death characteristics of whole-genome genes were obtained. The top 1% of genes with the highest disulfide death characteristics across the entire genome were selected as potential regulators of disulfide death.

2. The method for predicting disulfide death regulatory genes based on large-scale expression map correlation according to claim 1, characterized in that, The large-scale human gene expression mapping data is transcriptome data, and the known disulfide death regulator data includes regulators that promote disulfide death and regulators that inhibit disulfide death.

3. The method for predicting disulfide death regulatory genes based on large-scale expression map correlation according to claim 2, characterized in that, Known disulfide mortality regulators include: SLC7A11, SLC3A2, RPN1, NCKAP1, CYFIP1, WASF2, ABI2, BRK1, RAC1, ATF4, NUBPL, NUDUFA11, LRPPRC, OXSM, NDUFS1, GYS1, G6PD, PGD, TALDO1, TKT, PRC1, SLC2A1, SLC2A2, SLC2A3, and SLC2A4.

4. The method for predicting disulfide death regulatory genes based on large-scale expression map correlation according to claim 1, characterized in that, After obtaining the expression residuals of the transcriptome gene expression matrix, datasets with fewer than 80 samples are removed, and the expression residuals are used to replace gene expression levels to obtain the whole genome gene expression residual matrix.

5. The method for predicting disulfide death regulatory genes based on large-scale expression map correlation according to claim 1, characterized in that, The Pearson correlation coefficient or Spearman correlation coefficient were used to perform expression correlation analysis.

6. The method for predicting disulfide death regulatory genes based on large-scale expression map correlation according to claim 1, characterized in that, Bonferroni correction was used to eliminate false discovery rates when calculating the expression correlations between each gene and the collected known disulfide death regulators.

7. A method for assessing disulfide mortality characteristics, characterized in that, include: The weighted average of the gene expression levels of the target disease sample, target tissue sample, or target organ sample and the disulfide death feature value obtained by the disulfide death regulatory gene prediction method according to any one of claims 1 to 6 is used to obtain the weighted disulfide death feature value of the target disease sample, target tissue sample, or target organ sample and normal tissue. By comparing the weighted disulfide mortality eigenvalues ​​of target disease samples, target tissue samples, or target organ samples with those of normal tissues, and by screening out cancer types with more than 3 tumor and normal samples that retain non-zero gene expression and corresponding disulfide mortality eigenvalues, the evaluation results of disulfide mortality eigenvalues ​​of target disease samples, target tissue samples, or target organ samples are obtained.

8. A prediction system for disulfide death regulatory genes based on large-scale expression map correlation, characterized in that, include: The acquisition module is used to acquire large-scale human gene expression map data and known disulfide death regulator data; The processing module is used to convert large-scale human gene expression map data into transcriptome gene expression matrix, and to calculate the expression residuals of transcriptome gene expression matrix using the PEER method to obtain whole-genome gene expression residual matrix. The analysis module is used to calculate the expression correlation between each gene and known disulfide death regulators using the genome-wide gene expression residual matrix; The prediction module is used to use the correlation of disulfide death regulators as the weighting factor of the weighted average correlation, and to obtain the disulfide death characteristics of the whole genome by calculating the weighted average correlation between the whole genome genes and the known disulfide death regulators. The top 1% of genes with the highest disulfide death characteristics across the entire genome were selected as potential regulators of disulfide death.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for predicting disulfide death regulatory genes based on large-scale expression map correlation as described in any one of claims 1 to 6.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for predicting disulfide death regulatory genes based on large-scale expression map correlation as described in any one of claims 1 to 6.

Images