Is there a good way to group probes from a methylation array (to get regions instead of single positions) and get methylation ratios for those regions?
I would say minfi/bumphunter is a pretty good strategy for identifying differentially methylated regions
The code may need to be updated for some programs, but you can find templates for running this type of analysis here:
https://sourceforge.net/projects/cohcap/files/Protocol_Exchange_Example.zip/download
http://www.nature.com/protocolexchange/protocols/2965#/introduction
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
I know there are tools for differential methylation, but that's not really what I want. I don't want to group samples and find significantly different regions. I am just trying to collapse the full probe list for each sample. Regions are more meaningful than probes.
Exactly - bumphunter (in minfi), RnBeads, COHCAP, and IMA all define differentially methylated regions (in addition to sites, as an upstream step).
If you only have one sample, you can also use COHCAP to identify differentially methylated regions in that sample, but I would recommend the group analysis (with any of the 4 tools listed above) if you have multiple samples and want to looks for differences between groups.
I don't want differentially methylated regions which are going to change every time. I just want a constant set of regions that I can apply to any sample and then figure out later whether it's differentially methylated or not.
In relation to this question, @Charles Warden, how would you compare DMRs among 3 different conditions? I understand bumphunter output for 2 groups but I do not see the logic for 3 groups.
Thank you!
EDIT: I opened a new post describing my doubts more deeply.