1
$\begingroup$

I understand GSEA/GSVA can take microarray-like expression matrix such as the output of voom() or vst(). However, I have a question on how we can also use svaseq() variables, correct the output from voom() or vst(), and apply GSVA on it.

I assume this is possible since I have read papers showing PCA after correction by SVA (such as this one and this one in their supplementary figure), but my statistical and coding knowledge does not allow me to recreate what they do. Is there a known pipeline that can do this

Edit: to clarify, I would like to generate a normalised matrix (e.g. rlog, or out put of voom from limma, or vst from DESeq2) that is adjusted for batch effect. ComBat, for example, can provide what I need. However, since I am doing my DESeq2 DGE analysis with svaseq(), I want to adjust it with svaseq variables. I am wondering if there are pipelines that can use svaseq vairables (or any varaibles) and adjust the value in the normalised expression matrix.

$\endgroup$
0

1 Answer 1

1
$\begingroup$

Use removeBatchEffects from limma. The input counts should be on log scale, so vst and rlog are ok. The covariate argument can take the surrogate matrix from svaseq.

https://rdrr.io/bioc/limma/man/removeBatchEffect.html

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.