Fig. 3

Subunit types based on PRGs in PCa and their immune microenvironment. A Consensus clustering matrix for k = 2. B The results of K-M analysis indicated that cluster 1 had a significantly lower tumor progression-free survival than cluster 2 (log-rank test, p = 0.005). C Principal component analysis (PCA) showed that these PRGs could well divide TCGA cohort into two distinct clusters. D Heatmap with different clinical features revealed that PRGs had differential expression in different clusters. E Immune cell infiltration of different clusters based on "CIBERSORT" algorithm (Wilcoxon test, *P < 0.05; **P < 0.01; ***P < 0.001). F The heatmap shows the biological pathway to pyroptosis-related clusters by gene set variation analysis (GSVA) (|log2FC|> 0.2 and adjusted p < 0.05)