Fig. 2
Total genes with at least one high-confidence transcription start site (TSS) after random subsampling. For each library individually (Panel a) or for all libraries within each experiment combined (Panel b), from 10 to 70 % of reads were randomly subsampled in increments of 10 %. The number of reads (in millions) subsampled for each library/experiment was plotted (x-axis). The reads in each subsample were then mapped to the genome, and the total number of genes with at least one high-confidence TSS peak was computed (y-axis)