Hi all! I'm running a WGCNA to correlate gene modules with physiological measurements. My experimental design has two temperatures and three genotypes, and I am interested in the interaction between genotype and temperature. I went through the official tutorial and then the refinebio tutorial, and I'm a bit hung up on the modeling part. When I ran the differential expression analysis, I used the model ~ genotype + temperature + genotype:temperature. I ran WGCNA just fine, and then in the refinebio tutorial you run a linear model on each module. I used the same design, with the interaction term, but then the results are just a big table with a single p-value per row- does this mean the p value is for the whole model?
I could of course run each genotype separately and compare them manually, but I am particularly interested in the interaction term. Is this possible? Does it make sense to run the model I want, with both main effects and the interaction term? Should I be using glm instead? I've attached a plot of the top module retuned by the model with the interaction term- it looks like there is a genotype specific temperature effect (first two letters are genotype, C or H is the temperature) so maybe I'm just second guessing? Any insights would be greatly appreciated! Thanks!
