MultiBamSummary for correlating ChIPseq datasets

Question

[brisbio]

I want to be able to determine if the enrichment of a specific histone modification is correlated with the enrichment in a specific TF binding in genomic regions that these both overlap. Having looked through examples I see you can use multiBamSummary and plotCorrelation to see if you have correlation between read coverages. Would this be a good package to use to answer my scientific question?

Answer 1

[Devon Ryan]

You may have to play around with the bin size since the TF site will presumably be quite small compared to the histone mark, but in general this seems like a reasonable way of attacking the question.

[brisbio]

Thank you. I will definitely have a play around with the bin size. I noticed in a separate post you commented that it is probably better to use multibigwigsummary as I can first normalise to input. So I think I will continue that way!

Answer 2

[Rory Stark]

Another tool for this is the Bioconductor package DiffBind, particularly if you have replicates. This would allow you to compute a set of consensus TF binding sites, centered around the point of highest pileup, and calculate correlations with the histone mark (or find places where it systematically changes in different conditions). The vignette gives a useful overview of how to do this straightforwardly.