Hi Biostars,
I have multiple murine whole genome sequencing samples which show a non-uniform readcount distribution along the genome, all following the same pattern, example attached. Has anyone be facing this pattern in readcount distribution before and may have an idea what could be causative? (i dont see patterns in standard QC parameters)
Thanks for any input on this!
Update on GC bias:
GC content was proposed as a reason, so I plotted mean GC in reference genome along with the readcount distribution. There is no clear correlation, maybe a mild high GC - low coverage, but it’s very subtle. Thus, I guess GC bias is not the (major) cause of the fluctuating coverage. Does anyone have a different idea?
Sequencing coverage by short reads depends on GC content (https://pubmed.ncbi.nlm.nih.gov/22323520/). Maybe, this coverage distribution reflects distribution of GC content along chromosomes?
Also, if you removed reads with low mapping quality, this may have reduced coverage in repetitive regions.
I had a quick look at overall GC content, which looked ok on a first look, but I'll dive deeper into GC, thanks for the hint!
Thank you for your comment! I was using HMMcopy (https://bioconductor.org/packages/release/bioc/vignettes/HMMcopy/inst/doc/HMMcopy.pdf) which is correcting for GC content and Mappability, but I'll anyway check if the pattern relates to the GC content pattern.
GC normalisation is only for small-scale (~1kbp) differences in GC content. GC bias does not explain arm-level copy number biases. That fact you're seeing higher coverage toward all centromeres can't be explained by GC. The abrupt changes (e.g. chr12/chr13) can be explained by genomic rearrangements, but the gradual arm-level changes (e.g chr2) looks more like a sequence artifacts than real biology.
This agrees with my current picture of the problem. Thank you for your opinion!
Log in to answer this question.
