I have three groups let's say C, L, and T, each with three replicates. I did comparison between C vs L, C vs T, and L vs T to find the Deferentially expressed genes in each compassion. I used RUVs package to find the DE genes. I want to draw a heatmap of these differentially expressed genes, but I have different set of DE genes in each case. I have following two questions on this:

1. What values should I use to generate the heat map? Is it raw counts OR the fitted values after the GLMfit function?
2. How to draw the heatmap for if their are different set of genes in each case?

Thanks!

This isn't really a question about how to use Bioconductor software, but instead is a question about the 'best' way to present data. As the analyst, it's your job to figure that out, based on the rationale for the experiment, and what you want to highlight in your paper or poster. I would imagine that most people would use the logCPM values for the heatmap, because they are more comparable than the raw counts would be, but otherwise these are questions you should be asking yourself or your collaborators.