I'm working through the edgeR user guide and would like to understand the reasoning behind using
glmQFTest() to detect DE genes, specifically within the oral carcinoma Case Study (p40 onward). I see from the user guide and also edgeR glmLRT with test="F"? and other sources that for cases where the number of replicates is small,
glmQFTest() is preferred for the reasons stated. However, in that particular Case Study there appears to be three biological replicates in each group, yet
glmLRT() is used. Is there a reason for this? Is the choice related to the paired design (i.e. only one sample in each group within the pair) or something else? I note that in section 3.4.1 on p35 in the example covering the ideas behind paired designs, there is a similar design and
glmQFTest() is used, I just can't work out why!
Any help understanding this would be very much appreciated.