I have a question regarding multiple samples in a paired dataset. I tried to conduct differential expression analysis between samples accessed from different sites of paired patients. For example, patient 1 may have a skin sample and two brain samples collected in this dataset.

      Patient.ID   Accession.Site
1           1          Skin      
2           1          Skin      
3           1          Brain       
4           2          Brain    
5           2          Brain    
6           2          Skin

The design formula I used is ~ Patient.ID + Accession.Site

This may be trivial to ask but I want to have a clear idea whether DESeq2 model will average those multiple samples from the same site for the same patient.

Thanks, Jeff

Yes, you could think about what DESeq2 would do here is similar to averaging over the replicates of skin for patient 1, etc. It's not exactly the arithmetic mean because we use a GLM, and because the LFC for accession site is fit over all patients.