Can I normalize several samples with DESeq2 and then do pairwise comparisons?

Question

[bioinfo]

Hello,

I have 5 groups of samples each comprised of 3 samples. I did pairwise comparisons between the groups (eg group1 vs group2, group1 vs group 3 etc). However, I printed the normalized counts and for example the normalized counts for group 1 in the first comparison is different to the ones from the second comparison.

Is there a way to add all the samples in one DESeq2 analysis so they can all be normalized at the same time and then do pairwise comparisons?

Thank you

Answer 1

[rpolicastro]

Yes, you can and should add all the samples to the same DESeq2 object and specify the contrasts on that joint object. Adding all of the samples to one object will usually improve the dispersion estimates for each gene.

The DESeq2 manual details working with multiple groups so you can follow their vignettes.