Figures and data

Multi-omics consensus clustering classification. (A) Calculation of optimal number of subtypes based on cluster prediction index and gap-statistics. (B) Heatmap of the consistency of the integration of the results of the 10 typing algorithms. (C-D) Significant multiomics characterization and prognostic differences between subtypes. (E-G) Validation of consistency of typing results and prognostic value in an external cohort.

Differences in molecular characterizaiion of subtypes. (A) Differences in mutation landscapes of subtypes. (B-C) Genomic characterization from TMB and FGA of subtypes. (D-E) Functional emichment of up and down-regulated differential genes between subtypes. (F) Concordance between subtypes and patients’ pathologic staging.

Phenotypic featmes of subtypes that may be associated with tumor development. (A) Landscape of subtype Th1E immune infiltration. (B) Differences in subtype responsiveness to different drugs. (C) Comparison of immune cell infiltration levels in subtypes ofTME. (D) Comparison ofhepatocellular carcinoma-related signaling pathways and biological process activity in subtypes.

Building prognostic prediction models through machine learning. (A) 101 machine learning algorithms to build screening optimal prognostic models(TCGA1=training set,TCGA2=internal validation set). (B) Regression coefficients of 5 genes obtained in stepwise Cox regression.

Comparison and refinement of prognostic prediction models. (A) Comparison of predictive performance of prognostic models with some prognostic models published in the last 5 years. (B) Constructing a nomogram incorporating patient clinical characteristics. (C-D) Evaluating model performance based on calibration curve and decision curve analysis.

ScRNA-seq cell s01iing and annotation. (A-C) Cellular subpopulation delineation and annotation based on cellular characterization genes. (D-E) Visualization of the proportions of each cell subgroup. (F) Differentially expressed genes by cellular subpopulations.

Phenotypic characterization of cellular subpopulation. (A-B) Functional enrichrnent analysis of differentially expressed genes within cell subpopulation. (C) Analysis of the metabolic landscape of cell subpopulations. (D) Cellular polymeric communication network. (E) The contribution of signal strength in both incoming and outgoing signals shapes the landscape of cellular subgroups. (F-G) Analysis of signal transmission into and out of cellular subgroups.

Exploring the characteristics of two subtypes of malignant cells at the scRNA-seq level. (A) Mapping of malignant cells in the cell clustering results. (B) Enrichment analysis of differential genes between different subtypes of malignant cells. (C) Metabolic level analysis of different subtypes of malignant cells. (D) Annotation of the cellular subpopulation signaling landscape after malignant cell annotation. (E-G) Analysis of signal input-output patterns in cellular subpopulation. (H) Differential levels of the MIF signaling pathway in cellular subpopulation outputs.

Signaling modulation pattern of CSl subtype malignant cells on other cells.