Dmrseq parameters for short regions

[jckhearn]

Hi,

I am running dmrseq on an organism with much shorter and sparser runs of CpG methylation (100s of bp) on average than vertebrates.

At the moment I am comparing default parameters with 'laxer' ones, detailed below:

cutoff = 0.025
maxPerms = 25
minNumRegion = 3
bpSpan = 300
minInSpan = 5

Is this a sound approach or can you suggest better parameter settings to explore for this analysis?

Thanks,
Jack

[kdkorthauer]

Hi Jack,

That sounds like a reasonable approach to detecting differential methylation where CpGs are sparser and the correlation is more local. The main parameters that will have an influence here are the 'bpSpan' and 'minInSpan'. I would suggest comparing a couple of different settings where these have been lowered (just as you proposed).

Best,
Keegan