Fig. 3

Establishment of a gene signature related to glycolysis and hypoxia. A Identification of DEGs in HNSCC. Using |log2FC|> 1 and FDR < 0.05, a total of 3391 DEGs in HNSCC were identified. B 3391 DEGs were used to construct the WGCNA network to identify non-grey modules. C Construction of co-expression modules related to key cancer hallmarks. The blue module was identified as the module with higher correlation with glycolysis and hypoxia (r > 0.3, P < 0.0001). D Using univariate Cox analysis, 97 candidates related to the prognosis of patients with HNSCC were identified from the genes of the blue module (P < 0.05). E Using random forest, the top 10 genes with the highest gene importance were screened out. F The gene combination with a relatively small number of genes and a relatively significant P value was selected from the multiple combinations of 10 genes to construct a survival prediction model. DEGs, differentially expressed genes. FDR, false discovery rate. WGCNA, weighted gene co-expression network analysis