Fig. 3
From: Deep dissection of stemness-related hierarchies in hepatocellular carcinoma
Identification of prognostic stemness-related signatures (A) Random forest algorithm to calculate the top genes' relative importance. B The top 10 genes were subjected to 1023 (210–1) permutations and log-rank tests, and the top 8 genes were selected for model construction. C Analysis of SRscores among different subtypes in TCGA-LIHC data, ***p < 0. 001. D Analysis of survival differences between high and low SRscores groups in TCGA-LIHC data. E, F ICGC-LIRI-JP data to validate the fitness of the model. G, H Time-dependent ROC analysis showing the specificity of SRscores in TCGA-LIHC (training set) and ICGC-LIRI-JP (validation set). I Pie charts showing the cardinality test for clinicopathological factors and stemness subgroups for each group of the stemness score