Quantifying Peak Height Difference in ATAC-seq

Question

[jjstevens]

Hi, everyone. How would I go about quantifying the peak height difference of a particular gene between two a wild type and a mutant in ATAC-seq? Any help would be much appreciated, as I'm very new to this!

Answer 1

[ATpoint]

Isn't that just a differential analysis of the read counts, so the amount of reads intersecting each peak? You can do that with frameworks like edgeR; DESeq2 or limma, or specialized wrappers such as csaw and DiffBind, all in R at Bioconductor.

[Marco Pannone]

I just want to add that in case choosing to proceed with DiffBind, to pay attention to how the algorithm defines overlapping (consensus) peaks among the two sample groups. By default, DiffBind considers peaks overlapping if they simply overlap for at least 1bp of their length, which, in my opinion, is absurd.

It seems pretty much a gold standard procedure that two peaks can be defined as overlapping if they overlap reciprocally for 50% of their length.

I think it is more reasonable to create your own overlapping peaks dataset, like using bedtools intersect, then feeding it into DiffBind pipeline.