Figure 3

Saturation, sensitivity and correlation of peak calling with decreasing cell number. (a) Saturation of peak calling as reads are randomly discarded. Peaks were called using only unique non-duplicated reads. (b) Overlap of called peaks in the different datasets with benchmark dataset peaks. Inset diagram defines examples of full or partial peak overlap, with the upper bar in each case representing the benchmark. Colours as in panel a. (c) Coverage of benchmark peaks by peaks in other datasets. Colours as in panel a. Inset shows examples of coverage, with upper bar in each case representing the benchmark. (d) Correlation of peak heights between benchmark and new sample datasets. Spearman correlation coefficients (ρ) are given. Only peaks overlapping a benchmark peak were included in this analysis. The number of reads in a given peak was normalised to the total number of reads (uniquely mapping non-duplicated) in the sample.