Fig. 5

Chromatin structural features of differentially active domains. a Schematic of the comparisons: 7 mRNA expression comparisons between normal and tumor samples from the same tissues were matched with Hi-C datasets from normal and tumor cells of the corresponding tissue. DADo was run on chromatin domains determined from both Hi-C datasets for a total of 14 comparisons. b Ratio of shared boundaries between domains inferred from normal and tumor cells from the same tissue (green), only differentially active domains determined from these comparisons (orange), and all chromatin domains inferred from all Hi-C datasets that we analyzed (gray). c Distribution of the ratios of shared boundaries between domains inferred from normal and tumor cell Hi-C datasets. The distribution was built by 1000 random sampling of n = 446 domains. The observed ratio obtained for the 446 differentially active domains is shown by the red dashed line (76.5%) obtaining an empirical p value = 0.11 from the expected distribution of ratios. d Fold-changes between the number of significant differentially active domains changing from A to B (blue) or B to A (red) compartment in the normal vs. tumor comparisons and non-significant domains, as determined by DADo. e Distribution of the domain rank difference of matching chromatin domains in normal and tumor Hi-C datasets for all chromatin domains (black line) and differentially active domains only (green density distribution). f Percentages of all chromatin domains (top pie chart) and significant differentially active domains (bottom pie chart) that are in the A (red) or B (blue) compartments. g Fold-changes between the number of significant differentially active domains and non-significant domains, as determined by DADo, that are in each of the 8 chromatin sub-compartments inferred by Calder. h Distributions of mean intra-domain gene expression correlations (left) and mean intra-domain gene expression levels (right) in each of the 8 chromatin sub-compartments inferred by Calder