We compared MACS with three other publicly available ChIP-Seq peak finding methods, ChIPSeq Peak Finder [8], FindPeaks [11] and QuEST [12]. To compare their prediction specificity, we swapped the ChIP and control samples, and calculated the FDR of each algorithm as Number of control peaks / Number of ChIP peaks using the same parameters for ChIP and control. For FoxA1 and NRSF ChIP-Seq (an FDR for CTCF is not available due to the lack of control), MACS consistently gave fewer false positives than the other three methods (Figure 2a,b).
