Specificity and sensitivity of MeDiChISeq peak predictions compared with other algorithms. A manually curated ChIP-seq benchmark dataset for MAX, SRF and NRSF  was used to assess the percentage of true site recovery by the indicated peak calling algorithms relative to the false discovery rate (FDR). For the two upper panels no background control sample was used during peak calling. The two lower panels show the same analysis but with considering the corresponding background control dataset in the analysis. Note that PeakRanger performance is not illustrated in the upper panels, as this algorithm does not perform IP binding site assessment without background control. In cases where the tracings overlap, an arrow indicates the point of maximal recovery.