I have two groups of patients whose gene expression is expected to change differently over time. In one group, the change in gene expression is expected to be greater or at least different, but individual patients are also expected to have different baseline levels of gene expression as well. Therefore, I'm interested the interaction of time, and patient, and not in individual patient effects.
In my DESeq2 model I use ~Patient * TimePt, and I would like to do a contrast in which I combine the patient.latetimepoint interactions in one group vs the patient.latetimepoint interactions in the other group. For this, I think I need to use lfcShrink with type="ashr" and a contrast argument, and would also need something along the lines of the listValues argument, but it doesn't appear that lfcShrink has such a thing. However, it does appear that ashr::estimate_mixprop can take a "weights" argument, and that that may be what I want to use. Is this correct, or am I barking up the wrong tree?
Thanks for your help,