I'm using limma in Bioconductor for differential expression analysis of an Affymetrix data set. The goal is to determine the genes that are differentially regulated with the overexpression of an insert gene via a viral vector. The design is such that we have four sets of arrays- vector only, no vector, overexpressed gene (via vector), no inserted gene (but still has vector). This is so that we can account for the expression of the vector itself if there ends up being variation between the viral expression with the insert and without the insert.

It seems like I should be able to setup the contrasts similarly to 'factorial designs'- so my contrasts now are: vecvsnovec = vector - vectorcontrol, insertvsno.insert = insert - no.insert, diffinsertvs_vec = (insert - no.insert) - (vector - vectorcontrol)

Does this make any sense? Now I'm almost thinking that I could probably just do insert vs no insert and leave it at that, to avoid the possibility of overcompensating. At least with the contrast design above, I could just use that one contrast.

Sorry for the vague descriptions, but any input is appreciated.