Different CpG calls when using different regions of inclusion for Methyldackel extract

[MoniDR]

Hi,

We've been using methyldackel for the analysis of our EM-seq libraries and I've noticed a bit of an odd behaviour from methyldackel. Specifically, when I specify different regions of inclusion (i.e. bed file A and bed file B, where file B should be a subset of file A), methyldackel returns slightly different regions in the final bedGraph files. In some cases a CpG regions not present in the bedGraph generated with -l fileA seems to be present in the bedGraph generated with -l fileB. I'm using the following parameters:

varFrac = 0.5
OppDepth = 10
minCov = 10
OT = 6,140,6,140
OB = 5,143,5,143
-l fileA/fileB

(OT/OB boundaries have been empirically defined using Methyldackel mbias).

Any idea as to why this is?

Thank you so much in advance fr your help!
Monica