Consensus Peak Shifts after dba_count() and Default Peak Width in DiffBind

Question

[Siqi]

Hi all,

After running dba_object <- dba(sampleSheet = samples, minOverlap = 2), the first consensus peak starts at chr1:10024 and ends at chr1:10326. However, after this then running dba_object <- dba.count(dba_object, minOverlap = 2), the same consensus peak shifts slightly, starting at chr1:9905 and ending at chr1:10305.

I understand that these two represent the same consensus peak, but I’m curious about why the start and end positions show this minor shift. How are the consensus peak locations and summit loci defined during this process?

Additionally, I noticed that after running dba.count, the width of all peaks is set to 400 bp. Is this a default setting in DiffBind, and if so, can it be customized?

Thanks!