Entering edit mode
I am trying to compare two different conditions (with 2 replicates each) using Diffbind. For my Diffbind output, I'm getting 0 counts for one of the conditions and high counts for the other condition. This doesn't seem to match my normalized browser tracks. There's clearly a peak around the chromosome position, but Diffbind is calculating it as 0 counts. There's definitely still a change in peak intensity between the two conditions, but it doesn't seem like it's reduced to 0. I was wondering if I'm doing something wrong or if I'm not understanding how the counts are calculated?
Could you show us the script you are using for the DiffBind analysis?
I notice you are using
bUseSummarizeOverlaps=FALSE. Do you have a particular reason for doing so? Do you get the same behavior when you use the defaultbUseSummarizeOverlaps=TRUE?Also, can you try running it with
summits=0and see if you still are getting zero counts?Hi Rory. I discovered that the culprit were my input files. I had another question though. When using
summits=250, some of my peaks aren't 500 bp. Although most of them are 500 bp, some of them are an odd width. Is there a reason for that? Thanks!Yes, this is possible.
In the first step, the summits are computed for all merged peaks in the consensus peakset. Next, the peaks are re-centered around this summit, so will all be of width
summits*2+1. If any of these newly re-centered peaks overlap, they will be merged, and hence will be wider thansummits*2+1bp.It is possible to force recaluation of the summits, re-centering and merging, until all of the peaks have the same width:
with output:
Thank you!
-redundant entry removed-
-redundant entry removed-