Deseq2 : Paired analysis or otherwise

Question

[IM]

Dear Community members,

I want to conduct a DESeq2 analysis on 20 samples from 11 donors on a data from a longitudinal study regarding exposure to a certain agent. Each donor has 2 samples except 2 donors (samples discarded due to poor quality). 6 donors have one sample each in both groups of comparison, whereas 3 donors' samples are all in one group. For such an analysis, would a paired analysis be recommended?

Answer 1

[ATpoint]

Yes, paired analysis is almost certainly your only chance since with human data variance between donors is typically too high to run unpaired analysis at that sample size. You will need to exclude unpaired donors as they don't add for any inference. Of course try unpaired with all samples as well but I doubt it will yield anything substantial.

[IM]

Thank you for your reply, ATpoint! Would you recommend that samples from donors 6,8, and 9 also be removed along with unpaired samples of donors 4 and 5?

[ATpoint]

In a paired analysis these samples do not add, so maybe remove them.