Hi,

I'm working with DNA methylation data obtained from the EPIC chip. I've previously used the R package Bumphunter for differentially methylated regions (DMRs). In my current project, I would like to specify a list of specific regions for bumphunter to analyse. In other words, I'm looking for a way to "fix" the regions in advance, and then tell Bumphunter to look at those specific regions - in stead of Bumphunter analysing DMRs genome wide.

Does anyone have any experience or know if it is possible to tweak bumphunter commands to restrict such regions?

I'm very grateful for any response and help to get me a step closer to solve this :)

Thanks for reading!

There is a 'cluster' argument to bumphunter that you can use to specify the regions. Not sure how that would work in practice if you wanted to ignore other regions. I would imagine you would need to subset your GenomicRatioSet to include only those regions as well. In other words, the cluster just tells bumphunter which CpGs are in a given cluster, not that certain clusters should be ignored.

Restricting to certain regions may adversely affect the inference provided by the p-values as well (which aren't necessarily easily interpreted as it stands). So caveat emptor?

I see what you're saying . Yup, seems like restricting such regions would not be optimal for what I have in mind. I'll give the 'cluster' argument a try, though not sure how that will work since I'd like to specify at least 1000 regions. I'll take a closer look at it, and maybe I need to rethink my line of reasoning .

Thank you for your input!