Evaluation of suspect list filtering on SEACR peak calling. Left: Comparison of peaks called using SEACR against a negative control on relaxed and stringent modes for the beta-catenin A replicate, showing that suspect list filtering of BAM files before peak calling leads to a smaller number of called peaks which have an increased average signal profile within the peaks and an increased percentage of peaks which contain expected TCF/LEF motifs (middle). The increase in background signal in the filtered and more stringent sets in the average profile (left) is due to an increase in peaks called in high-signal regions, not suspect list filtering (see Supplementary Fig. 1 D). Right: Comparison of peaks called by SEACR on MYC and MAX datasets from [ 1 ] before and after suspect list filtering, showing a decrease in peak number accompanied by an increase in average signal within the peak regions.
