I've a RNAseq counts dataset of 32 samples belonging to two groups: group1 (16 samples) and group2 (16 samples). Each group contains 2 subgroups: subgroup1 (8 samples), and subgroup2 (8 samples). Now, by using glmLRT test I want to find deferentially expressed genes between subgroups inside group1, and do the same inside group2. What I've tried so far is the following:
condition <-factor(c(rep("group1subgroup1", 8), rep("group1subgroup2", 8), rep("group2subgroup1", 8), rep("group2subgroup2", 8))) design <- model.matrix(~ 0 + condition) colnames(design) <- levels(condition) # make contrasts contrasts <- makeContrasts(gp1sub1.vs.gp1sub2 = group1subgroup1 - group1subgroup2, gp2sub1.vs.gp2sub2 = group2subgroup1 - group2subgroup2, levels = design) # define dge list dge <- DGEList(counts=rnaseq.counts) disp <- estimateGLMCommonDisp(dge, design) disp <- estimateGLMTrendedDisp(disp, design) disp <- estimateGLMTagwiseDisp(disp, design) # fit the model fit <- glmFit(disp, design) # LRT test for contrast 1: between subgroup1 and subgroup2 inside group1 lrt.1 <- glmLRT(fit, contrast = contrasts[,1])
I've obtained positive and negative log2 FC for the genes.
Interpretation: A gene with a positive log2FC means that gene is over-expressed in subgroup2, and a gene with a negative log2FC means the gene is under-expressed in subgroup2.
Am I right ?