Fig. 4: Polymer simulations of cohesin- and CTCF-dependent TAD dynamics.

a, Parameters used to model loop extrusion in polymer simulations. Three parameters characterizing cohesin dynamics (density, residence time and motor speed) were varied systematically, while the CTCF genome-wide residence time and occupancy of CTCF sites were held constant. b, CTCF residence times used for simulations at the four genomic regions. The color and orientation of arrowheads indicate CTCF site orientation, with height reflecting CTCF residence time. c,d, Contact maps from polymer simulations of the L1 TAD for different combinations of cohesin residence time, density and motor speed. In c, the motor speed is fixed to 1 kb s⁻¹; in d, the cohesin density and residence time are fixed to 12 Mb⁻¹ and 22 min, respectively. In b–d, dashed black lines indicate the location of TAD anchors (or of fluorescence reporters for the half TAD locus). e, Simulated 3D anchor–anchor distance time series of the L1 TAD. Simulations include random localization errors and photobleaching consistent with experiments, and reporter–anchor separations. Simulations assume a cohesin density of 12 Mb⁻¹, a residence time of 22 min and a motor speed of 0.25 kb s⁻¹. Panels c–e were created in part with BioRender.com.