Fig. 2: RCMC finely resolves dynamically changing focal looping interactions.

a, Contact map visualization of RCMC data at the Klf1 (bin size: 1.6 kb, left; 1 kb, zoomed-in view) and Dag1 (bin size: 3.2 kb, left; 2 kb, zoomed-in view) loci across the M-to-G1 transition, with Hi-C data¹⁸ (left) and the superset of loops (right) shown below the diagonal. Genomic annotations and ChIP data (stage-specific and asynchronous) are shown at the bottom. b,c, Histograms of loop interaction distances (b) and the number of interactions formed by each annotated anchor (c). d, Pie chart of annotated loop anchors by their genomic identity, determined by chromatin features within 1 kb of anchor sites. Promoters were identified as annotated TSSs ± 2 kb, enhancers as nonpromoter regions with overlapping H3K4me1 and H3K27ac ChIP-seq peaks and CTCF/RAD21 as nonpromoter and nonenhancer sites with overlapping CTCF and RAD21 ChIP-seq peaks. Anchors with multiple overlapping genomic features were hierarchically classified into a single classification, with promoters taking precedence, then enhancers and finally CTCF/RAD21. Anchors designated as ‘other’ do not overlap promoters, enhancers or CTCF/RAD21. e, Pie chart of annotated loops by the genomic identity assigned in d, with P designating promoters, E designating enhancers and CTCF designating CTCF/RAD21. f, Swarm plot of the number of interactions formed by each annotated anchor, separated by the genomic categories shown in d. g, Plots of individual P–P interaction strengths in the PM (left) and LG1 (right) conditions, plotted against the strengths in the AT condition (x axes). Strengths are calculated as the integrated observed loop signal divided by the expected background signal from local P(s) curves (‘observed over expected’). ‘Exclusive’ P–P loops are shown; 186 P–P loops that overlap with CTCF at one or both anchors are removed and all subsequent loop pileups and quantifications of strength by loop identity similarly omit loops meeting both CRE and CTCF/RAD21 loop types.