Hello all,

I am trying to use DiffBind on my ATAC seq data. However, the code in the manual uses ChIP seq data. Would you advise me on what to do for a learner? Thank you so much!

Update: maybe I got a good answer.

Within DiffBind, the main thing is to change the value of the summits parameter when calling dba.count(). The default value of 200 results in 401bp regions being considered. Setting summits=75 results in uniform peak widths of 151bp which works pretty well for ATAC data.

While you can see peaks closer to 200bp depending on nucleosome positioning, using 151bp widths is still OK as it is not essential to capture the entire open region when doing a differential analysis. Using a narrower region that is more likely to be enriched (less background) can more reliably identify differential enrichment between ATAC peaks. After these are identified, it depends on what biological question you are addressing; in most cases, you don't need the precise boundaries of the differential open regions, only what they are proximal to (or overlap with). In cases where the precise boundaries are important, you can include a downstream step to adjust the re-centred peak based on the original peak calls.