Multi-IP normalization. This figure shows CHANCE output for the multi-IP normalization module. (a-d) CHANCE produces a summary statement (a-b), a pairwise sample differential enrichment matrix (c-d), and a graphical representation of the normalization process. The graphical representation gives the same type of plot as in IP strength estimation for each IP sample, as well as the consensus of the IP samples; see Materials and methods. The summary statement quantifies the graphical representation by giving the statistical significance of the difference of each sample from the consensus. The differential enrichment matrix computes the percentage of the genome differentially enriched between all pairs of samples, using the same technique for IP-Input comparison used in IP strength estimation; see Materials and methods. (a,c,e) Multi-IP normalization of H3K4me1, H3K4me2, H3K4me3, and H3K36me3 in human embryonic stem cells (H1 HESCs), from the Broad ENCODE data. (b,d,f) The capacity of CHANCE multi-IP normalization to detect batch effects. The clustering of technical replicates (denoted by 1 and 2) for each biological replicate (denoted by A and B) seen in (f) is quantified in the pairwise differential enrichment matrix (d), which shows a statistically insignificant percentage of the genome differentially enriched between replicates but a non-negligible percentage of the genome differentially enriched between batches.
Diaz et al. Genome Biology 2012 13:R98 doi:10.1186/gb-2012-13-10-r98