I suppose our question now is: does this warrant changing the default behaviour of

bedtools genomecov

within ChIP-seq, and

bamCoverage

within CUT&RUN to result in more realistic background levels? PS: As an aside, since the underlying bedGraph and bigWig files are not wrong at the individual position level, one can hope that peak callers do the right thing and NOT USE any binning prior to peal calling… Does someone want to check?