Supplementary Figure 7: Flowchart describing the generation of the models and the virtual Hi-C data.

(a,b) 4C data (a) were translated into distance restraints that were added to the rest of the restraints (b). (c) The genome was represented as concatenated beads of different size that represented 20 fragments. The size was proportional to the sum of the read counts. (d) Models were optimized, starting from randomized bead positions. (e) After 50,000 iterations, we selected the 200 models with the best score. These models were clustered on the basis of their RMSD. (f,g) The models from the most populated mirror image cluster were superimposed (f) and the virtual Hi-C heat map was generated by calculating the mean distance between all the beads from all the models (g).