Hi,

In my dataset, the number of samples from each individual are different, something like this:

I want to compare active and inactive status, but as you can see, patient 1,2,3 has 1 samples in each status, whereas patient 4,5,6,7,8 contribute multiple samples in each status.

In this scenario, can I design my Deseq model as:

~ status + Patient.ID

to eliminate the unequal variance issue?

Thanks! Leran

In DESeq2 we assume a dispersion per gene, so anyway we never model unequal variance/dispersion conditional on X.

You can just use our standard design of ~patient + status.