Hello,

I'm facing a quite complex experimental design with 2 factors (Group and Time) and with paired samples. If the number of patients is the same for each group, i can use the trick proposed by EdgeR in "Comparisons both between and within subjects" section and then by DESeq2 in “Model matrix not full rank” section. But is there a way to deal with unbalanced groups ?

Imagine the following design :

If I try to do the same thing than with balanced group, i get a “Model matrix not full rank” error.

Best

You need to remove the column that represents the interaction between group A and patient 4. This is discussed towards the end of that section you reference "Note that, if you have unbalanced numbers of individuals..."