one further finding: I think I may have found some further clues… I have just done a test with a

-bg

and the

--bga

files:

bigWigSummary H3K27me3.bigWig 4 95027300 95028500 20
n/a     n/a     n/a     n/a     n/a     0.035752        0.0674382       0.0701722       0.0355933       0.0273398       0.0273398       0.0273398       0.0273398       0.0273398       0.0273398       0.0273398       n/a n/a      n/a     0.0273398

bigWigSummary H3K27me3_BGA.bigWig 4 95027300 95028500 20
0       0       0       0       0       0.0232388       0.0674382       0.0701722       0.0314408       0.00455663      0.0273398       0.0273398       0.0273398       0.0273398       0.0273398       0.0250615       0   00       0.0159482

The region should have some regions with 0 coverage, some with 2-3 reads, and few bins with just 1 read. While the values generally agree well, the non-zero bin file (the above one) are consistently higher when a bin only covers a fraction of a read, where it seems to assign a minimum value of 0.0273398 (which is the $SCALE_FACTOR) . The

--bga

version can be considerably lower, e.g. 0.00455 in the middle. This argues against a mere IGV artifact…$