Screening of the Pyroptosis-associated Gene METTL7B in Sepsis Prognosis Based on Gene Weighted Co-expression Network and Machine Learning Algorithms and Its Expression Validation
Objective To screen pyroptosis-associated genes in sepsis prognosis using weighted gene co-expression network analysis(WGCNA)and machine learning algorithms,and to validate their expression in sepsis models.Methods The gene datasets GSE95233,GSE65624,and the single-cell transcriptional sequencing dataset GSE167363 were downloaded from the Integrated Gene Expression Omnibus(GEO)database,with GSE95233 serving as the training set and GSE65624 and GSE167363 as validation sets.Differentially expressed genes(DEGs)were then identified by comparing the sepsis group with the normal control group.Using pyroptosis-associated genes,unsupervised clustering was performed with the ConsensusClusterPlus R package.DEGs were then screened based on the clustering groups within the dataset.Additionally,WGCNA was employed to identify the module eigengenes most associated with sepsis prognosis.By combining sepsis-related DEGs,pyroptosis-associated DEGs,and module eigengenes,candidate genes associated with sepsis prognosis were identified through Venn diagram.The key gene METTL7B related to pyroptosis in sepsis prognosis was screened using Least Absolute Shrinkage and Selection Operator(Lasso)regression,Random Forest(RF),and Support Vector Machine(SVM)analysis.Diagnostic and prognostic receiver operating characteristic(ROC)curves and survival curve analyses were performed for the key gene METTL7B in the training set GSE95233 and the validation set GSE65624.Finally,based on the high-throughput sequencing data from the dataset GSE167363,the expression of METTL7B in septic macrophages was analyzed and demonstrated,and the expression of METTL7B in sepsis modeling macrophages was verified by RT-PCR.Results The study identified 203 sepsis-related DEGs,689 pyroptosis-associated DEGs,and 891 module eigengenes.14 candidate genes associated with sepsis prognosis were obtained through Venn diagram.Subsequently,the 14 overlapping genes were screened using Lasso regression,SVM,and RF analysis.The Lasso regression model retained six feature genes with non-zero coefficients.The RF algorithm indicated that the highest accuracy of the model was achieved when the number of feature genes was five.The SVM analysis revealed that the minimum 10-fold cross-validation error occurred when the number of feature genes was 4.Finally,METTL7B was identified as a potential key gene related to pyroptosis in sepsis prognosis through Venn diagram.In the training set GSE95233,the AUC values for METTL7B diagnosis and prognosis were 0.990 and 0.702,respectively.In the validation set GSE65624,the diagnostic AUC was 0.939.Additionally,in this dataset,septic patients with high expression of METTL7B had a higher survival rate.The expression of METTL7B in macrophages was found by single cell transcriptional sequencing analysis of GSE167363.The elevated expression of METTL7B in macrophages of sepsis was further verified in vitro(P＜0.05).Conclusion METTL7B,a key pyroptosis-associated gene in sepsis prognosis,screened through WGCNA and machine learning algorithms,may potentially serve as a prognostic marker for sepsis.
NETL
NSTL
万方数据
