Determining tight and loose paired regions in the PnM genome

A close examination of the PS track revealed two important features: (i) the fly genome is divided into regions that demonstrate consistently high, relatively similar, values of PS, followed by extended regions where PS dips into lower values, (ii) switching between high- and low-PS regions seemed to occur often around insulating boundaries.

We used the PS and insulating boundaries to determine the precision of pairing in the genome, and examine the variation of pairing more closely, in addition to the internal organization of tight and loose pairing. First, we divided the genome into regions between pairs of consecutive boundaries (as detected in reference-mapped, i.e., not allele-resolved Hi-C data). Second, we classified each of these regions as either tight or loosely paired, depending on the number loosely paired bins in that region. Because the dips in the PS track tend to be gradual, with PS being noticeably low only further away from the boundaries, we considered the cutoff of 25% of loosely paired bins per region to be sufficient to call the whole region as loosely paired. Finally, we noticed that this method occasionally split single loosely paired regions into a few smaller ones, presumably due to false positive boundary calls. To mitigate this issue, we merged consecutive loosely and tightly paired regions into larger ones if the boundary bin between them was classified the same way. We then used the detected regions of loose and tight pairing to calculate scaling curves P_cis(s)^loose, P_cis(s)^tight, P_thom(s)^loose, and P_thom(s)^tight. We calculated these curves using the same method as for the genome-wide P_cis(s) and P_thom(s), but only considering pairs of loci within the same region. See supplementary Note ¹ for a detailed discussion.