I'm trying to perform a differential expression analysis using DESeq2 and I have a question about normalisation factors. I have been provided with a set of technical confounders (GC means, insert size, primer index and date) and I used to add them to my design (e.g.,
"~Covariate + Technical Confounder + condition"). However, looking better at the DESeq2 vignette, I noticed that I can use sample/gene dependent normalisation factors (as created for instance by EDASeq from my technical confounders) and use them instead of the
estimateSizeFactor function. What is the difference between the two approaches? Which one produces the best (more reliable) result?
Thanks in advance for your help,