Fig. 2

Differential editing patterns are shared among cancer types yet distinct from differential gene expression. a Rank-rank hypergeometric overlap (RRHO) map of RNA editing across pairs of cancer types. Each heatmap (for two cancer types) represents the matrix of log10-transformed adjusted p values that indicate the extent of overlap in two gene lists at each possible pair of ranks. For an individual cancer type, genes were ranked by the signed significance of RNA editing differences (M-E). Genes with higher editing in the M phenotype are at lower ranks, while those with higher editing levels in E tumors are at higher ranks. Higher pixel darkness indicates stronger enrichment of overlapping genes within the rank thresholds given by the x and y coordinates. The step size between ranks was 30 genes. b RRHO map of editing and gene expression within each cancer type. Each heatmap contains log10-transformed adjusted p values of hypergeometric overlap between genes ranked by editing differences (x-axis) and genes ranked by expression differences (y-axis) in a single cancer type. Similar to ranking genes by differential editing, genes were ranked by the signed significance of expression differences, such that genes at lower ranks have higher expression in M tumors, while genes at higher ranks have higher expression in the E phenotype. The step size between ranks was 30 genes. c Significance of enrichment of gene ontology (GO) terms in differentially edited genes of each cancer type represented by point size (log10-transformed adjusted p value). Terms significantly enriched in at least two cancer types are shown. Check mark on the right indicates terms that were also significantly enriched in differentially expressed genes in at least two cancer types. Text color indicates category of biological relevance