I am using edgeR to try to construct a model to compare multiple conditions from some sequencing data. I have three growth conditions (A, B, C) and replicate samples for each condition at time 0 (T0) and after so many generations (T1). The edgeR manual was quite helpful in constructing a model to compare the individual growth conditions (T1 vs T0 for A, T1 vs T0 for B, T1 vs T0 for C) to identify differentially expressed genes for each condition. However, I am now struggling with how to construct a design/model.matrix to find genes that are unique and significant only in a single growth condition (only significant in A T1 vs T0 compared to B T1 vs T0 or C T1 vs T0) or significant only in two of the conditions (significant in B T1 vs T0 and C T1 vs T0 but not in A T1 vs T0).
The manual shows that I can find differences between two comparisons using the following construct:
> AvsB.contrast = makeContrasts(A.vs.B = (A_T1-A_T0) - (B_T1-B_T0), levels = design)
But I am a bit confused how to use this to compare between three conditions. One contrast I've tried is:
> AvsBvsC.contrast = makeContrasts(A.vs.B = (A_T1-A_T0) - (B_T1-B_T0) - (C_T1-C_T0), levels = design)
But I am not sure if that is giving me those genes that are unique ONLY in condition A and not in condition B or C.
Based on other reading on this site (which is great) I've found some examples and I've also tried this contrast:
> A.contrast<-makeContrasts(contrasts=c("(A_T1-A_T0) - (B_T1-B_T0)","(A_T1-A_T0) - (C_T1-C_T0)", levels=design)
But am again unsure if that is giving me the output I want - genes that are significant differentially expressed ONLY in condition A and not conditions B or C.
I also am not sure how to go about finding genes that are unique ONLY in conditions B and C but not A (or B and C individually).
Any help would be greatly appreciated. Thank you in advance!