Hi all,

I want to run sva and include the results in the design passed to deseq2 looking for degs.

Sva takes normailzed counts, but deseq2 takes raw counts and normalises also.

So if I proceed, the data will be normalise twice.

Is it ok to normalize twice?

Should I remove surrogate variables and de-normalize before passing to deseq2?

Should i run deseq2 but prevent it normalizing again?

I realise sva is not strictly necessary here, but I'm just using a simple example to identify the issue.

Regards, Kenneth

We have an SVA and RUV example in the workflow.

Briefly it does not double normalize. The factors are estimated on scaled data and so DESeq2 will use these to remove variance after accounting for size factors through its own method.