Fig. 2

Comparison of CLIP peak calling methods. a Barplot showing the Jaccard similarity index of the peaks identified by each computational method (shown in the legend) from replicate experiments. For all RBPs, except PUM2, data were available from two distinct cell lines, HepG2 and K562. b Distribution of Jaccard indices of peak overlaps extracted by individual tools (color code as in a) from replicate samples. c Heatmap showing the Jaccard similarity index of the peaks identified by two distinct methods for a given protein in a given cell line (in percentages, averaged over two replicate experiments). The methods are shown in order in the x-axis and the same order (RCRUNCH, CLIPper, omniCLIP, and PureCLIP) is used in each block (corresponding to one protein and cell line) on the y-axis. The average similarity of a method with any other method on all datasets is shown at the bottom. d Barplot showing the running times of peak calling steps for RCRUNCH, omniCLIP, and PureCLIP. Error bars show the standard deviations from the 2 replicate runs