Fig. 6
From: STARRPeaker: uniform processing and accurate identification of STARR-seq active regions
Comparison of peaks using an external dataset for a HepG2 or b K562 cell lines. Peaks identified from STARRPeaker as well as BasicSTARRseq and MACS2 were compared against a published dataset. For a fair comparison, all peaks were centered at the summit, uniformly thresholded using P value < 0.001, and 20,000 peaks were randomly drawn from peaks identified by each peak caller using the recommended settings. The fraction of overlap was computed for each replicate. We used the total number of peaks in each dataset as the denominator. We considered it an overlap when at least 50% of peaks intersected each other