Extended Data Fig. 1: Patient cohort and cluster information.

a, Pie charts showing the composition of cancer types in our cohort. HCC, hepatocellular carcinoma; ICC, intrahepatic cholangiocarcinoma; CHC, combined hepatocellular and cholangiocarcinoma; HH, hepatic hemangioma; ASC, adenosquamous carcinoma; SAR, sarcomatoid carcinoma; SLC, secondary liver cancer. CRC_M, liver metastasis from colorectal cancer, PAN_M, liver metastasis from pancreatic cancer, LYM_M, liver metastasis from lymphoma, GAS_M, liver metastasis from gastric cancer, BRC_M, liver metastasis from breast cancer. b, UMAP plots showing the distribution of patients, cancer types, viruses and liver cirrhosis states. Dots represent individual cells. PB, peripheral blood; AL, adjacent liver; HBV, hepatitis B virus, HCV, hepatitis C virus, NBNC, double negative of HBV and HCV. c, UMAP plots showing expression of canonical marker genes of major cell populations including T cells (CD3D, CD8A, FOXP3), NK cells (NKG7), B cells (CD79A), macrophages (CD68), neutrophils (CSF3R), dendritic cells (CLEC10A), mast cells (TPSAB1), fibroblasts (COL1A1), endothelial cells (VWF), and epithelial cells (EPCAM). d, Stacked barplot showing the distribution of major cell types in each sample. e, UMAP plots showing the distribution of cell identities for tumour cells and TIME cells. Tumour cells were further coloured by patient, cancer type, virus, and cirrhosis. f, CNV profiles inferred from scRNA-seq data for each cell and from matched bulk exome data in the sample A014_HCC. g, Boxplots showing hepatic scores and biliary epithelial scores in tumour (n = 193,877 cells) and TIME cells (n = 898,295 cells). Cells are from 124 patients. h, Boxplots showing hepatic scores and biliary epithelial scores in tumour cells of different PLC subtypes (HCC, n = 96,211 cells from 79 cases, ICC, n = 52,345 cells from 25 cases, CHC, n = 15,493 cells from 7 cases). Cells are from 111 patients. i, Pie charts showing the patient number (top) and cell number (bottom) of our study and published single cell studies for PLC. Colours represent different studies. j, Stacked barplot showing proportions of major cell populations among different studies. Colours represent major cell populations. In g-h, n denotes individual cells. Two-sided Wilcoxon rank-sum test is used. For boxplots, centre line shows median, box limits indicate upper and lower quartiles, and whiskers extend 1.5 times the interquartile range, while data beyond the end of the whiskers are outlying points that are plotted individually. ***, P < 0.001.