Fig. 2

Personalized immune profile evaluation. a The t-SNE plot shows the immune microenvironment of patients from single-cell transcriptome sequencing data using the comboSC workflow. The different colors represent the cell types annotated by comboSC. b The pie chart shows the proportion of cell groups in the tumor ecosystem and the population size of each cell group. c The heatmap shows the marker genes and their expression patterns in identified immune cells, malignant cells, and stromal cells. d A schematic description of tumor heterogeneity used in comboSC. e The proportion of CR, PR, and NR samples in 63 samples. f–h ROC curves of comboSC immune score from Tres model in f TCGA-SKCM, g Fraietta-CLL, and h GSE123814 datasets. H, M, and L in the figure indicate CR, PR, and NR samples, respectively. ROC curves and AUROC values for H–L, H-M, and M-L indicate the classification performance of CR versus NR, CR versus PR, and PR versus NR using comboSC immune scores, respectively. i The bar plot shows the precision of CR in top immune score samples for each dataset in the green left bar. The red right bar shows the threshold of top immune score samples in each dataset, which is defined as the proportion of CR samples