paired analysis : return data matrix with indvidual/subject effects eliminated.
Entering edit mode
Last seen 14 days ago


I'm doing methylation analysis on samples that are paired because they come from the same individual i.e. subject, and I'd like to get the data matrix with that effect eliminated or minimised because it's quite big in methylation studies.

There is good information on how to account for between-subject effects in models such as that found in the limma user's guide section 9.4.1, and in other places, but there is less about how to render the data matrix with those effects eliminated (so, for example, PCA - prcomp() could be run on it and so we would see how the subject's twin samples cluster together alot less).

Functions like sva's Combat() and RUVSeq's RUVs() do do this, it seems a bit of hack to use them for this aspect (RUVs() wants counts not beta values and ComBat() is finding singularities (too much correlation perhaps).

Any helpful comments from other users who have also run such analyses, especially in methylation studies?

Many thanks in advance!

RUVSeq sva DESeq2 • 89 views
Entering edit mode
Last seen 8 hours ago
United States

See removeBatchEffect in the limma package.

Login before adding your answer.

Traffic: 234 users visited in the last hour
Help About
Access RSS

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6