I'm trying to prevent Diffbind from merging overlapping peaks in my bed file (TSS file with exactly 2000 bp widths) by setting DBA$config$mergeOverlap to a high number. But, it's still merging overlapping peaks as if DBA$config$mergeOverlap isn't working. I'm not sure if I'm using it incorrectly.

samples <- read.csv("TSS.csv")

tamoxifen <- dba(sampleSheet="TSS.csv")
tamoxifen$config$mergeOverlap <- 5000

tamoxifen.counted <- dba.count(tamoxifen, summits=F, bUseSummarizeOverlaps=T, score=DBA_SCORE_NORMALIZED, filter=0)
counts <- dba.peakset(tamoxifen.counted, bRetrieve=T)
normalized_counts <- as.data.frame(counts)

I'm still getting an output with peaks that are obviously being merged with other peaks.

I'm taking a closer look at how this is working and I should have a fix soon.

In the meantime, the work-around is to add a second call to dba() after you have set the mergeOverlap config value, which will trigger a recalculation of the merged peaks using the supplied parameter:

tamoxifen <- dba(sampleSheet="TSS.csv")
tamoxifen$config$mergeOverlap <- 5000
tamoxifen <- dba(tamoxifen)

tamoxifen.counted <- dba.count(tamoxifen, summits=F, bUseSummarizeOverlaps=T, score=DBA_SCORE_NORMALIZED, filter=0)