Fig. 3: Specific interactions are sufficient to drive chromatin folding in Drosophila.

a The chromatin polymer ensemble is constructed sequentially, with the addition of one bead at a time. We make use of an optimized sampling distribution at each step to improve sampling efficiency, which is dynamically adjusted after adding each bead. b In our model, Hi-C contact probability corresponds to the proportion of polymer chains that satisfy the ligation threshold. c Illustration of simulation results for a 1 Mb region (chr2L: 11.0–12.0 Mb) of S2R+ cells at 2 kb resolution. Heat maps from left to right represent Hi-C propensities, simulated contact probabilities using all, specific, and non-specific interactions, respectively. d Pearson correlation coefficients of the simulated contact probabilities and Hi-C propensities in 10 regions of different genomic lengths. The number of beads ranges from 100 to 1000. e\(\mathrm{log}\,\)-\(\mathrm{log}\,\) scaling curves of contact probabilities with genomic distances (bin) derived from the original Hi-C data, simulated ensembles using specific, all, and non-specific interactions. f Distance distributions of the two anchors of the loop shown in c. Loop anchors correspond to bead No. 170 and bead No. 190. *** represents two-sided Wilcoxon rank sum test p-value ≤ 0.001. g Height maps of contact probability of the loop interaction inside the polycomb domain shown in c. h Constructed 3D conformations of chromosome X using specific interactions at 5 kb resolution. (Top) Simulated Hi-C heat map, with measured Hi-C propensities at the bottom and simulated contact probabilities at the top. (Bottom) Zoomed-in heat map of a 2.5 Mb region. Pearson correlation coefficient r is 0.94, distance-adjusted correlation \(r^{\prime}\) is 0.64. i Visualization of two examples of 3D conformations of chromosome X using PyMOL. Source data are provided as a Source Data file.