I have been scratching my head as to how I can create a design matrix to find which interaction is significant. My goal is easiest explained with some plots.
For this first one, I want to use limma to indicate that depending on time point, there is an expression difference in IPCs (among potential others). And ideally, limma would also tell me that Radial Glial for instance has no significant change between timepoints. Limma would be applied after a pseudobulking pipeline that I already have worked out.
I'd like to do something similar for one more type of comparison. Here, timepoint has been substituted for genotype. This is even more tricky, it looks like IPCs have a change in the HOM, but against what? The HET, the WT, or both? What about testing HET vs WT? It quickly becomes a lot of comparisons considering we also have cell-type.
Can anyone suggest the proper way to go about this, is limma even a suitable tool? Also should both these questions be addressed with one design matrix or should there be two and limma would be run twice essentially?
Thank you in advance for your help.
How many time points are there? Limma users guide 9.6 walks through some good examples for both few and many time points. Would the
decideTests()function maybe be able to help with what you want to accomplish, also?