DESeq2 count normalization clarification

Question

[poweroscar]

Hello, This is just to clarify something I read.

I've been using public data sets, getting gene counts of samples for DE analysis. And of course, you need to normalize the count. I've been following the steps of DESeq2 on this page here for normalization. The steps are clear, but when it comes to "normalize all samples", does it mean it normalize all samples of the same condition? or all samples regardless of conditions?

More specifically, it's regarding the pseudo-reference sample (row-wise geometric mean) of step 1 in the said page. If I have 3 conditions, do I get 3 pseudo-ref or only 1?

Thank you.

Answer 1

[ATpoint]

Normalization runs on all samples in the dds object regardless of the design. See also this excellent StatQuest explanation: