Fig. 5
From: Integrative epigenetic taxonomy of primary prostate cancer
Validation and characterization of three subtypes. a On the left: unsupervised hierarchical clustering of 285 differentially expressed genes between the three integrative clusters. The sample classification according to integrative cluster analysis is indicated below the branching. On the right: unsupervised hierarchical clustering on the TCGA cohort using the genes differentially expressed across the three clusters of the Porto cohort. The genes (rows) in the heatmap are ordered the same as for the Porto cohort. Color scale: red indicates high expression an blue low expression (z-score). b Heatmap of copy number alterations (CNAs) of 88 Porto samples (left) and TCGA samples (right). Samples from Porto and TCGA cohort are ordered the same as in Figs. 4a and 5a, respectively. Red and blue represent copy number gains and losses, respectively. c CNA burden as fraction of the genome that is copy number altered in the Porto cohort (left) and TCGA cohort (right). Boxplot: median values with interquartile range. p-Values were calculated using the Wilcoxon test. d Boxplot (median values with interquartile range) showing the number of genes with a mutation for the three clusters in the TCGA cohort. p-Values were calculated using the Wilcoxon test. e Association of mutation with the three clusters identified in the TCGA cohort. Only genes with significant differential enrichment among the clusters are shown (FDR < 0.2). Percentages on the left show the mutation frequency. f p-Values for top gene sets enriched (FDR < 0.2) from the MSigDB collection of “curated” gene sets in any of the three clusters, represented in a radar plot for Porto (left) and TCGA cohort (right). p-Value for each cluster is indicated with a line with the corresponding color
