Hi,

I am trying to analyze my 16S data using DESeq2, and I would like some advice on the correct formula I should use.

I have before/after data for two different treatments; also, I have 2 age classes. This is a sample of the table:

Patient Time    Treatment   AgeClass
p1      Before       A           C
p1      After        A           C
p2      Before       A           M
p2      After        A           M
p3      Before       B           C
p3      After        B           C
p4      Before       B           M
p4      After        B           M

As you can see, Treatment has two levels (A and B). I want to look at the effect of the two treatments, but also taking into account the age class of the patient. I was wondering if the following formula is correct:

~AgeClass + Treatment + Treatment:Patient + Treatment:Time

Should I make any change to the formula?

Thank you for your kind attention, Max

My first note is that I'm not sure DESeq2 is appropriate for microbiome datasets necessarily. There was a recent review paper posted to bioRxiv so maybe you can read that to determine if it's appropriate for your data.

https://www.biorxiv.org/content/10.1101/2020.01.15.907964v1

Otherwise, this is a common question on Bioconductor. If you control for patient, you should not also add patient characteristics to the design. This is unnecessary and will produce a linear dependence in the design.