Supplementary Figure 3: Comparison with 3C-based assays.
From: High-resolution genetic mapping of putative causal interactions between regions of open chromatin
(a), Classification of Hi-C annotation. Causal interactions were classified into (1) outside TAD: two peaks located in an unannotated TAD region, (2) one-side TAD: one of two peaks located in an annotated TAD, (3) across TADs: two peaks span two different annotated TADs, (4) within TAD: two peaks located inside the same TAD annotation, (5) across inner TADs: two peaks spanning one or more TAD boundaries within a TAD, and (6) loop anchor; two peaks, both located in a Hi-C loop anchor region. (b), The number of causal interactions overlapping with the six different annotations shown in a. (c), Enrichment of causal interactions overlapping with the six different annotations shown in a. To compute the odds ratio (OR), n = 15,884,515 peak pairs (peak distance > 35 kb) were used. (d), QQ plots of RASQUAL P values for allele-specific chromatin accessibility analysis stratified by the Hi-C annotation in a. (e), Enrichment OR with 95% confidence interval of the same effect direction in different 3C-based annotations. The number of causal peak pairs to compute the OR for each category was n = 3,060, 3,060, 3,060, 3,060, 3,060, 3,060, 11,534, 11,534, 11,534, 3,060, 11,534 and 11,534 from top to bottom. The bottom two annotations show the effect direction is less concordant when two peaks span TAD boundaries or insulators (CTCF peaks). (f), Estimated peak-pair-level prior probability excluding peak pairs separated by short or intermediate distances (1, 3 or 5 kb). (g), Number of mapped causal interactions with estimated prior probabilities in f. The left four bars show the total sum of posterior probabilities and the right four bars are based on the high-confidence set with PPCjk > 0.5.
