I need to find Differentially Methylated Regions and I was about to use the tool comb-p. In the tool you need to specify the distance between consecutive cpgs to be part of a region. I didn't find online a way to decide the parameters "distance", but beside the problem of the choice of the parameter, I was wondering how a DMR "should" be in terms of width (in probes) and how many cpgs should contains. For example, is it correct to have large regions (e.g. 2000 probes) with inside like 10 cpgs, or it would be better to have smaller regions (e.g. 500 probes) with 20 cpgs? I know that might not exist a specific answer, but do I have to decide the parameter to be more oriented on the first example or the second one?
What's the relation DMR width/number of cpgs? There is a minimum number of cpgs that a region should contains so that i have to filter the DMRs?
Any help on the interpretation, or any input for some paper/book where i can find this information is appreciated.