caQTL signals associated with multiple, coordinated peaks. (A) caQTL mapping strategy testing variants within 1 Mb of ATAC peak centers. A variant may affect the accessibility of a distal peak if it affects accessibility of the overlapping proximal “driver” peak, and the proximal peak affects the distal “response” peak. (B) Number of coordinated caPeak sets containing different numbers of caPeaks. (C) Distribution of genomic width covered by peaks in the coordinated sets. (D) Coordinated peaks at ARPP21. The predicted driver peak is indicated with the purple arrow, and the ARPP21 promoter peak in the coordinated set is indicated with the orange arrow. The ATAC signal tracks represent read depth–normalized ATAC signal averaged across individuals for the three genotypes of caQTL lead variant rs6784162 (purple arrow), which is the lead variant in the 1 kb and 1 Mb caQTL analyses. (E) Coordinated peaks at SORT1 with predicted driver peak (purple arrow) and one of the SORT1 promoter peaks (orange arrow). Read depth–normalized ATAC signal tracks averaged by genotype for rs7528419, which is the lead variant in the 1 kb and 1 Mb caQTL analyses. The previously described functional variant at this locus, rs12740374, is located in the same peak. (F) Overlap of driver peaks predicted by proxy overlap and PHM. (G) Enrichment of TF ChIP-seq binding sites from ENCODE in driver peaks relative to response peaks. Odds ratios and 95% confidence intervals. All TFs are significant at a P-value threshold of 2.9 × 10−3.
