Hi all, I am a little conflicted in determining which model to pick for DGE analysis. From what I read on an online forum, quasi-likelihood F test is more robust with GLM as it reflects the uncertainty of dispersion for the expression of each gene, and GLM is preferred compared to the classic edgeR approach for >=2 factor design which is why I have chosen the use of these 2 functions in my R script for EdgeR DGE analysis. However, is there a more deterministic way to choose which model to use accurately?

Any input is greatly appreciated, thank you!

All the methods are ok but we recommend quasi-edgeR for a bulk RNA-seq analysis with replicates because it gives the most rigorous error rate control.