intersecting insulated regions

[sergpolly]

[sergpolly]

I've treated any set of continuous bins insulated <-10-5 (using <0 would change numbers by <1%) as a single "insulated genomic region", e,g, https://github.com/dekkerlab/4dn_jawg/issues/2 . I intersected sets of "insulated genomic regions" for corresponding replicates to estimate reproducibility, which is arguably better way of comparing these calls than using Pearson correlation coefficient.

• it's inherent nature of insulation potential (having lots of true zeroes between insulation peaks) that allows to do this analysis using simple <-10-5 (<0) thresholding
• Everything looks consistent: https://github.com/dekkerlab/4dn_jawg/issues/1
• counts are on the same order as # of dots ...
• "insulation calls" looks more saturated , as opposed to dots, i.e. we get similar number of calls even using combined datasets (R1 vs R1+R2).