I have a rather large dataset (100+ samples and growing) and I would like to examine the effect of modifying the minimum number of samples where a peak must occur to be included in my DBA object. As far as I understand, one can use the argument minOverlap in dba.count to achieve that, but this forces DiffBind to count the reads again -- which takes a significant amount of time and IO bandwidth. Is there a way to obtain the counts only in peaks with a given overlap without repeating the process?
I know the DBA object retains the raw counts even if they are normalized, which makes me believe that such a feature may exist.
Thanks in advance!