Fig. 2

Screening biomarkers. A Weighted gene co-expression network analysis (WGCNA) of genes. Branches are distinguished by different colors, representing seven distinct modules. B, C Evaluation of the scale-free topology criterion with R2 = 0.83 and the soft thresholding power = 7. D Correlation plot of WGCNA modules with trait attributes. Rows and columns represent modules and trait attributes, respectively. Red denotes a positive correlation, whereas blue denotes a negative correlation. E Significance of each module’s genes. F Venn diagram showing the intersection genes of DEGs and the turquoise module. G Cross-validation for choosing LASSO regression model tuning parameters. Three genes were examined, and dotted vertical lines were created at the optimal levels. H Screening genes using the SVM–RFE algorithm. The number of genes with the optimal accuracy is represented by the blue dots. I Venn diagram showing the discovery of significant genes using LASSO and SVM analyses