Fig. 2: SIDISH identifies malignant high-risk cell population and prognostic biomarkers in PDAC.

a UMAP clustering of major cell types, including two subtypes of ductal cells (ductal cell types 1 and 2), fibroblasts, T cells, acinar cells, B cells, stellate cells, endothelial cells, macrophages, and endocrine cells. b UMAP visualization of 3623 high-risk cells identified by SIDISH is shown in red, while the 38,363 background cells are depicted in gray. c Bar plot showing the distribution of high-risk cells across identified cell types. Type 2 ductal cells comprise the largest proportion of high-risk cells (55.8%), followed by type 1 ductal cells (16.4%), fibroblasts (7.9%), and T cells (5.9%). d Bar plot comparing the distribution of high-risk cells between tumor-only samples and combined tumor-control samples. e Volcano plot of differential gene expression analysis between high-risk and background cells. Upregulated genes are marked in pink, while downregulated genes are shown in blue. Adjusted P values were calculated using a two-tailed Wilcoxon rank-sum test. f Violin plots comparing the expression levels of key upregulated genes between high-risk and background cells, with the middle bar indicating the median expression for each cell subpopulation. P values comparing the expression levels between both cell groups were calculated using a two-tailed Wilcoxon rank-sum test. g Functional enrichment bar plots. The SIDISH-derived marker genes are enriched in biological processes and pathways related to PDAC progression and poor survival. These genes also show a strong association with PDAC-related disease terms. h, i Kaplan–Meier survival curves for two independent bulk validation datasets: GSE224564 (h) and GSE85916 (i). High-risk patients shown in pink show clear stratification from background patients shown in gray (P = 6.67 × 10−16 and P = 9.60 × 10−19, respectively), validating the clinical relevance of SIDISH-derived marker genes. P values were calculated using the two-tailed log-rank-sum test to compare survival curves between high-risk and background patient groups.