Fig. 4

Summary of single-cell sequencing data from HCC patients and subtypes of macrophages. A Of 32 samples with HCC single-celled sequencing data analysis, using the UMAP plot shows 11 cell clusters, colored by cell cluster. B Employing the CopyKAT algorithm to determine benign and malignant cells, most of the hepatocytes were inferred as malignant cells. C UMAP visualization of Scissor algorithm-identified Scissor-positive and Scissor-negative cells, Hepatocytes, Macrophages, and NK T cells are considered scissor-positive cells. D Heatmap shows differentially expressed genes (rows) identified by cNMF clustered by their expression across single cells (columns) from a representative patient. The gene clusters reveal intratumoral programs that are differentially expressed. The corresponding gene signatures are indicated (right). E Unbiased Clustering reveals six programs in HCC, Heatmap depicts pairwise correlations of 64 intratumoral programs derived from 32 tumor samples. F GSVA was used to compute the EMT pathway for spatial localization at the tumor-normal interface, with red representing the highest intensity of each SPOT expression and blue representing the lowest. G Using cytoTRACE to predict the differentiation potential and direction of macrophage subtypes; more.diff means higher differentiation potential, and as the end point of Slingshot, the less.diff is considered the starting point of Slingshot. H Using Slingshot to analyze macrophage subgroup trajectory of differentiation, relies on the results of cytoTRACE, S100A8+ Macro as the starting point, and the rest of the cell type as the destination. (cDC: conventional Dendritic cell)