Entering edit mode
Hello!
Since I do not have biological replicates I want to call DMRs with DSS. As far as I understand, first a basic smoothing is performed before using a beta-binominal GLM that takes biological and technical variance e.g. read-depth into account.
I have to admit, that I have not fully understand the theoretical background -- that is probably why I have to ask the following question: Prior analysis, should I filter out CpGs with low coverage or should I explicitly not filter any CpGs?
I thought I have read that DSS is ignoring CpGs with coverage < 5 but I can not find that reference no more. But that would be also a threshold I am considering to choose -- If necessary or recommended.
Any advice highly appreciated.
Best
Might be a bit late for the original poster :') but in case anyone else is wondering, the DSS user guide says:
Q: Do I need to filter out CpG sites with low or no coverage? A: No. DSS will take care of the coverage depth information. The depth will be factored into the statistical computation.
https://bioconductor.org/packages/release/bioc/vignettes/DSS/inst/doc/DSS.html