Fig. 5

Construction of 10 ferroptosis-related gene signature in the TCGA training set. (a) The optimal value of penalty lambda (λ) is selected in LASSO analysis. (b) LASSO coefficient profiles of 10 FRGs in OSCC. (c) The protein-protein interaction network among 10 candidate prognostic genes (red font) and 20 additional genes. The size of the node and the thickness of the connection line represent the interaction degree and the number of supporting pieces of evidence, respectively. The distribution of risk score (d) and survival time and status (e). The optimum cut-off value of the risk score is shown in (d) and used to divide each patient into different risk groups. (f) Heatmap of mRNA expression of the 10-prognostic FRGs. (g) K-M curves for the overall survival between low and high-risk OSCC patients. (h) Time-dependent ROC curves for the prognostic model for 1-, 3-, and 5-year OS of OSCC in the training set. (i) Time-dependent ROC curves for the risk score and clinical parameters in 1-year OS of OSCC in the training set. The AUC values represent the predictive performance of the gene signature and clinical risk factor