Hello,

This question is in regards to using DiffBind (or DESeq2) with multiple samples.

Currently, I am using DiffBind (version 2.8.0) to find differential peaks from ATAC-seq data. I have 7 different cell types with 1-3 replicates each (15 samples total). I'm looking to identify differential peaks among all 7 cell types in order to find peaks that are uniquely specific and significant to each cell type. It is unclear to me on how to set up the contrasts (or if it is necessary).

I would like to export this analysis in a CSV file, with rows as peaks, and columns indicating chrom, chromStart, chromStop, and p-value. So I would have one p-value for each peak corresponding to whether or not that peak is a differentially open region in one of the cell types.

Any suggestions are welcome. Thank you for your help.

Eliza

You can add a contrast for each cell type vs. everything else, so you'd have up to 7 contrasts. You can add each contrast as follows:

myDBA <- dba,contrast(myDBA, group1=!myDBA$masks$CellType1, group2=myDBA$masks$CellType1,
                             name1="Other Cell Types", name2="Cell Type 1")

You could do an "occupancy"-style analysis, where you form consensus peaksets for each cell type and overlap them to see which peaks are unique to a single cell type, but you wont have any statistical evidence to back this up.