Additional file 1: Figure S1. Identify the weighted value β that meets the law of scale-free networks. Figure S2. (A) Heatmap of the expression levels of 61 SHRGs in normal and tumor tissues. (B) Protein-protein interaction (PPI) network construction. (C) The main enriched entries for these genes. Figure S3. NMF rank survey. Figure S4. The mutation rates of the top fifteen most significantly mutated genes were significantly different between the Cluster 1 and Cluster 2 subgroups. Figure S5. The prognostic model could further differentiate patients with different clinical characteristics. Figure S6. Identification of GO and KEGG enrichment between high- and low-risk scores subgroups. Figure S7. The mutation rates of the top fifteen most significantly mutated genes were significantly different between high- and low-risk scores subgroups. Figure S8. The predictive significance of the prognostic model was verified in the nomogram. Figure S9. Top 16 most important tumor-sensitive drugs. Table S1. Clinical characteristics of HCC patients involved in the study. Table S2. The sequences of the qRT-PCR primers used in this study. Table S3. Immune cells with differences in abundance between Cluster1 and Cluster 2 by the four algorithms. Table S4. Immune cells with differences in abundance between high- and low-risk score groups by the four algorithms. Table S5. 77 tumor-sensitive drugs targeting tumor cell stemness.