Fig. 3: Upregulated PRC2⁺-CGI genes are characterized by high levels of cancer-type specificity and regulatory plasticity.

a Expression fold-change between tumor and nonmalignant samples, stratified by CGI promoter classes. Expression fold-change was calculated by DESeq2. p-values between two upregulated groups were determined by a two-sided t-test. p < 0.001***, p < 0.01**, p < 0.05*. The exact p-values are shown in Supplementary Data 4. Box plots indicate the median (middle line), 25th, 75th percentile (box) and 5th and 95th percentile (whiskers); the gene number (n) of each CGI class in each cancer type is listed in Supplementary Table 2. b Individual genes plotted for COAD, exemplary genes from Fig. 1c–e were highlighted. c Cancer type-restricted genes are identified based on expression fold-change between a specific cancer type (COAD in this example) versus all other cancer types. Fold-change is shown on the left, with all genes sorted by fold-change and those with ≥2 (“cancer-type-restricted genes”) shown in red. The 56 cancer-type-restricted genes for COAD are shown as a heatmap on the right. d The percentage of cancer-type-restricted genes from each gene class, shown by cancer type. e Plastic genes were defined as those assigned to the upregulated group in one cancer type, and hypermethylated in another. The percentage of plastic genes is the number of plastic genes in each cancer type divided by the total number of upregulated genes in that cancer type. f TCGA methylation and expression data are shown for a plastic gene (DKK1). g PCA analyses using expression values from each of the three CGI gene classes. The black circle represents squamous cancers (LUSC, HNSC, and a subset of BLCA) and the dark red circle represents gastrointestinal cancers (EAC, COAD and READ). In order to keep the scale consistent, extreme outliers were removed: 6 samples from left, 1 from middle, and 277 from right. h The average PCA distance ratio of inter-tumor versus intra-tumor samples for each class of CGI genes. Intra-tumor distance: the mean distance of all tumor pairs within the same cancer type; inter-tumor distance: the mean distance of all pairs in different cancer types.