CNVkit on WGBS data

[QLZhouBio]

Hi, I have some WGBS data and would like to check the copy number variant in these data. However, I am not sure whether CNVkit is suitable for WGBS data since the methylated C is converted to T in WGBS data. Will that affect the CG content hence affects the rolling median steps?

[kenji-yt]

@QLZhouBio I am pretty sure bisulfite conversion affects the relationship between GC content and coverage. This is because un-methylated regions will have very low GC content after conversion (1/2 original GC content if all C->T on both strands). I believe that reads will therefore be amplified by PCR according to their GC content post bisulfite conversion. Since PCR is the main source of GC bias (Speed and Benjamini 2012) it will affect coverage. For example in some of my own WGBS data there is no relationship between GC content and coverage. Normalization is therefore ineffective. The package ReadDepth accounts for this. They take the effective GC content of the bin by considering methylation status of each cytosine site.