Hello All

Is this the correct design matrix for the following experiment? The experiment has a Normal cell line (P) and a Mutant cell line (X) that were grown with decreasing nutrient levels, A, B, and C.  A is represent 100% nutrient for a particular component, B represent 50% , and C represents 0%.  The Normal cells (P) could only grow in A and B, while the Mutant cells (X) could be grown in A, B, and C.  The objective is of the study was to identify gene expression levels that changed (i.e., gene sensitive to the nutrient) that allowed the mutant to grow without the nutrient.  

Design Matrix:

celltype <- factor(samples$ct, levels = c("P", "X"))
trt <- factor(samples$trt, levels= c("A", "B", "C")) 
design <- model.matrix(~0+celltype+trt)
rownames(design) <- samples$file
design
colnames(design) 
#[1] "celltypeP"   "celltypeX" "trtB"         "trtC"

Thanks

The additive design won't account for interactions between celltype and trt, which would seem to be important if you want to identify mutant-specific responses to the nutrient level. I would suggest this:

grouping <- paste0(celltype, ".", trt)
grouping <- factor(grouping)
design <- model.matrix(~0 + grouping)
colnames(design) <- levels(grouping)

To identify relevant genes, you can test for genes that are DE in the mutant with respect to decreasing nutrient concentration (e.g., between X.C and X.B). This will pick up any genes that get turned on in the mutant when the concentration hits zero, and may be involved in mediating cell survival. You can't contrast this to normal cells, because they don't survive at zero, but you can do something equivalent with the concentrations that you do have:

con <- makeContrasts((X.B - X.A) - (P.B - P.A), levels=design)

This will identify genes that respond to decreasing concentrations in a different manner between normal and mutant cells, which may be involved in predisposing the mutants to survival (or the normal cells to death) when the concentration hits zero. Finally, you can do a straight up comparison between the genotype averages across all concentrations:

con <- makeContrasts((X.A + X.B + X.C)/3 - (P.B + P.A)/2, levels=design)

This will identify genes that are constitutively DE between genotypes and may be involved in driving mutant survival.

Thanks Aaron. I actually had another design where it is same as you have suggested. Please correct me if I am wrong. 

design <- model.matrix(~0+group)
rownames(design) <- samples$file
design
colnames(design)
#[1] "groupP_A"  "groupP_B"    "groupX_C"  "groupX_A" "groupX_B"

#For interaction

lrt <- glmLRT(glmfit, contrast = c(-1,1,0,-1,1))

For the second suggestion, could I still use contrast as:

lrt1 <- glmLRT(glmfit, contrast = c(-0.5,-0.5,0.33,0.33,0.33))

Would this give me the same answer too as (X.A+X.B+X.C)/3 - (P.B+P.A)/2 contrast?

Thanks