Hello,
I'm designing a new in vivo study wherein I will be topically treating the hind legs of mice with 1 of 3 drugs or a vehicle. To minimize the number of animals I would randomly assign each leg to 1 drug/vehicle. I will then be isolating the treated epidermis from each site after several weeks of treatment. I'll then subject the resulting RNA for RNA-seq and run DE analysis using DESeq2.
To account for differences due to the individual animal/subject I would like to use the following model:
design(dds) <- ~ subject + treatment
My question is whether or not I should design the study such that every animal has one (randomly assigned) leg receive the vehicle, and the other leg receive 1 of the 3 test drugs. Alternatively, can I simply randomize the vehicle and drugs together across each leg (ie, some animals will receive two different drugs but no vehicle). To me the first option would allow me to adequately account for differences due to subject while estimating the effect due to the drug. The flipside is that the first option would mean more animals, and many more "Vehicle" samples compared to any of the drug samples.
Any advice on the design of this experiment would be greatly appreciated!