I've got an expression data set with a wild type and a knock-down genotype both treated with two different dosages of a drug.
I'd like to get the genes that are more effected by treatment (ideally across dosages) in one genotype compared to the other. Since both the genotype and the treatment contribute to differences in expression I'd have to "correct" for the effect of the genotype.
Conceptually, does it come down to a variable intercept model where I look for a significant difference in slopes or how do I best model this? (should be possile in limma or would I need some kind of mixed-effects model using lme4 etc. ?)
How would I build the formula for the model in limma and how would I then extract the relevant p-values for the contrast mentioned above? (a code example with the variable names genotype & dosage would be awesome).