Fig. 3: Characteristics of epithelial cells within three cohorts.

a UMAP results of epithelial cells from all patients (16,415 cells from 14 samples). b Gene set enrichment analysis (GSEA) of epithelial cells from different cohorts. PZ, 3895 cells from 4 samples; TZ, 9049 cells from 6 samples; PTM, 3471 cells from 4 samples. The statistical significance of enrichment was evaluated using GSEA, with a nominal P < 0.05 and a false discovery rate (FDR) < 0.25 considered statistically significant. c Enrichment of different clusters in three cohorts. PZ, 3895 cells from 4 samples; TZ, 9049 cells from 6 samples; PTM, 3471 cells from 4 samples. The positions of the enriched (OR > 1.5) or depleted (OR < 0.5) clusters were shown in the UMAP plot through in situ staining. d The characteristic markers for cluster 0, 6, and 11. Cluster 0, 2808 cells from 14 samples; Cluster 6, 1126 cells from 14 samples; Cluster 11, 102 cells from 14 samples. e Cluster 6 signature for patient stratification in TCGA and CPGEA databases. High signature score patients in TCGA, 248; low signature score patients in TCGA, 249; High signature score patients in CPGEA, 60; low signature score patients in CPGEA, 60. Log-rank test for Kaplan–Meier survival analysis. f, g Expression of androgen response genes in the enriched or depleted clusters within PTM samples (3471 cells from 4 samples). Violin plots depict the kernel density of single-cell expression values. Median, 25th percentile, and 75th percentile are shown in the box plot; whiskers extending to 1.5 X IQR (interquartile range). h, i Dynamic network biomarkers (DNB) analysis for the most drastic changes across gene expression network of epithelial cells in trajectory I and II (16,415 cells from 14 samples). Two-side pairwise Wilcoxon signed-rank test with Benjamini–Hochberg for P value correction. *, ** and *** denoted P < 0.05, P < 0.01 and P < 0.001, respectively.