Fig. 1

Effects of data normalization. a Pie charts illustrating the changes in number of common and replicate-specific promoter-associated H3K4me3 peaks for HepG2 cell line datasets (GSM646364; GSM646365) before and after normalization (Blue: Rep1-specific peaks, Red: Rep2-specific peaks and Green: Peaks common between replicates). While mostly peaks overlap rate are conserved, some changes are observed post normalization in less enriched peaks, thus influencing peak calling thresholds. b An illustrative MA transformation plot shows the overall transition of RCI differences between replicates before and after normalization. The LOESS fit line (blue) shows the overall correction change after normalization. c Average RCI plots over annotated promoters (TSS with flanking regions of 1.5Kb) show that significant amplitude difference exists with peaks that are common between replicates (Blue: Rep1 and Red: Rep2). However, after normalization such amplitude differences are corrected and replicate-specific enrichments become more distinctive