Fig. 2: 3D interaction profiles discern three types of silent chromatin in HCT116.
From: Diverse silent chromatin states modulate genome compartmentalization and loop extrusion barriers
a, Two example regions illustrating the contrasting interaction profiles of B4 domains (left, chr2:3.5–3.6 Mb) and B0 domains (right, chr3:131–150 Mb) against A1, A2 and B1 in cis. The IPG labels are displayed as colored bars on the top and left margins (A1, red; A2, yellow; B0, green; B1, blue; B4, purple). Top, ChIP–seq tracks for HP1α, H3K9me2, H3K9me3 and H3K27me3. b, Heatmap of mean fold enrichment of ChIP–seq signal intensity for histone modifications, H2A.Z, and HP1α and HP1β proteins averaged over 50-kb bins in each interaction cluster (k = 8). c, Metaplots of B0, B1, B4 domains, rescaled to 25 bins and flanked by ±500 kb, displaying signal enrichment for ChIP–seq (H3K27me3, H3K9me2, H3K9me3, H2A.Z, HP1α/β/γ), Protect-seq and DNA methylation. d, E1–E2 scatter plots of 50-kb bins colored by ChIP–seq signal enrichment (H3K27me3, H3K9me2, H3K9me3) and ChromHMM state annotation. e, ROC curves assessing the prediction performance of individual 50-kb-aggregated functional tracks (ChIP–seq, Protect-seq) when treated as binary classifiers for B0, B1 or B4 loci. The discrimination parameter in each case is a simple binarization threshold on the entire signal track.
