Extended Data Figure 4: Insulation at globule boundaries.
From: Cohesin-dependent globules and heterochromatin shape 3D genome architecture in S. pombe
The degree to which a locus displays decreased contact frequency between, or is insulated from, regions separated by that locus can be directly quantified from the corrected Hi-C contact map. Here we use Rj(s), the relative frequency of contacts occurring over a bin j at a distance s. Negative values of relative contact frequency, Rj(s), are indicative of insulation at a given locus. Rj(s) at a given distance s is calculated from a region within a rectangular band of a Hi-C contact map rotated by 45°. a, Diagram illustrating the concept of the insulation plot. At the location of the cohesin binding peak, interactions between two adjacent globules are less frequent (blue stripe). Within the globule domain, contact probability is high (red stripe). b, Relative contact probability around a cohesin peak as a function of insulation distance averaged over all cohesin peaks. Average insulation is examined by calculating the relative contact probability around cohesin peaks. Relative contact probability around the cohesin peak is depleted up to ∼50–100 kb, indicative of insulation at peaks of local cohesin enrichment at these scales. c, Relative contact probability averaged from 20–50 kb around positions of each cohesin peak (positions obtained in wild type were assayed in rad21-K1). d, Mean number of cohesin peaks as a function of distance from boundaries. Psc3 peaks are highly enriched at the boundary in wild type. e, The negative correlation between cohesin and relative contact frequency Rj(s) in wild type indicates that not only is insulation observed at peaks of cohesin enrichment, but that the inverse relationship between the local enrichment of cohesin (Psc3) and the relative contact frequency holds genome-wide for data binned to 10 kb. This indicates that it is not just the presence or absence of a cohesin peak, but the local amount of cohesin protein in the chromatin fibre that may be important for boundary formation, as well as the strength of a given boundary. The negative correlation holds up to ∼100 kb in wild type. In rad21-K1, however, there is no appreciable correlation with Psc3 at any distance. This indicates that there is no clear relationship between the distribution of cohesin and local chromatin organization in this mutant.
