Hello,

I am interested in using diffbind to perform differential expression analysis on some ATAC data. However, all I have is counts table for peaks (no beds or bams) and a meta table with the sample names and conditions they belong to. Is there a way to use diffbind from here and is there any benefit over simply using DESeq2 or edgeR?

Thanks for the help in advance!

Fake example table

peakname condition1r1 condition1r2 condition2r1 condition2r2

peak1 19 21 29 31

peak2 34 30 15 17

You can import count data directly into DiffBind using the Counts column in the sample sheet. This is documented in the help page for dba.peakset(). It assume you have already generated a consensus peakset (same intervals for each sample) and counted the reads in each interval, so that you have a file or vector of counts for each sample.

The advantages of using DiffBind for this are mostly that you can use its reporting and plotting functions. The main limitation is that your experimental design need be pretty straightforward, otherwise you are better off using DESeq2 or edgeR.