Fig. 3From: Integrated machine learning identifies epithelial cell marker genes for improving outcomes and immunotherapy in prostate cancerPrognostic association of ECMGs classifications. A Consensus clustering matrix of 543 ECMGs categorized into two clusters (C1 = 179, C2 = 167). B 3D-PCA plot showing the distribution between C1 and C2. C Kaplan–Meier curves for biochemical recurrence-free survival (bRFS) in C1 and C2. D Violin plot comparing the copy number variation (CNV) frequency between C1 and C2. E Composition percentage of the two clusters in clinicopathological features, including age, Gleason score, pN stage, PSA, and pT stage. F Gene set variation analysis (GSVA) of the Gene Ontology in C1 and C2. G GSVA analysis of the Kyoto Encyclopedia of Genes and Genomes in C1 and C2. ***P < 0.001Back to article page