Fig. 6: Data-constrained computational model of DNA loops and TADs.

a Schematic of the Rouse polymer/loop extrusion model which includes physiological values of the dynamic properties of chromatin, Cohesin and CTCF and is further constrained by 3D DNA tracing data. Best-fit parameters are displayed (see also Supplementary Fig S7). b Genomic overview, 3D consensus representation, difference matrix as well as consensus representations from the labelled seriation groups from the simulated 100 kb region at Chr17:43.6-43.7 Mb. Examples of ground truth loops from the simulation are shown with grey lines indicating loops with one or two anchors outside the displayed region, while the blue lines indicate loops inside the region. The thickness and colour of the lines scales by their frequency in the simulations. c Pairwise Pearson’s correlations for individual experimental and simulated traces from Chr17: 43.6-43.7 Mb. Experimental data from two independent replicates per condition. d Cumulative pairwise distance distributions between the indicated genomic positions in Chr17: 43.6-43.7 Mb for the indicated simulation conditions. e Simulated and experimental consensus representations and median pairwise distance maps from a 1.6 Mb region on Chr2 and a 1.8 Mb region on Chr7 traced with 11 probes with ~150 kb spacing (see Supplementary Fig. S5) in control cells and corresponding simulations. Grey spheres indicate standard deviation of each position compared to consensus representation. f The same genomic regions as shown in (e) after WAPL was depleted by auxin treatment of HeLa WAPL-Halo-AID cells for 2 h (HeLa ΔWAPL), while in simulations Cohesin residence time was increased 20-fold. Number of traces as indicated in (e) and (f) from three (HeLa control) or two (HeLa ΔWAPL) independent experiments. Scale bars in (b) and (e) are 100 nm. Source data are provided as a Source Data file.