Fig. 3.

Case study on hepatocellular carcinoma. (A) The scRNA-seq of Hepatocellular carcinoma was obtained from Sharma et al. (2020); (B) UMAP for patientID distribution after Harmony batch correction; (C) UMAP for the cellular clustering; (D) Auto-resolution is performed by scDrug; (E) Sub-clustering for potential tumor cells from (C); (F) Gene set enrichment analysis for gene ontology annotation of a cell cluster; (G) An example of KM curve for survival analysis; (H) Heatmaps showing the potential drugs in the PRISM database predicted to inhibit cell growth by CaDRReS-Sc. Each cell represents values of predicted sensitivity score of the tumor cell clusters to the drugs; (I) Heatmap showing the optimal combined treatment of six drugs in the LINCS L1000 database to kill the greatest number of cell clusters according to the Premnas prediction.