Hi all

I am using TCGA data in order to find differentially expressed genes. The cohort I am using has just 41 normal samples, while the tumor subtypes I am working with have measurements ranging from 80 to 140 samples. Is EdgeR (or any other equivalent package) an option or I cannot get significant results from this kind of analysis? I know that group sizes are not required to be the same in EdgeR, but is a comparison between 41 and 141 samples reliable?

Thank you in advance.

R.

It is no problem at all. There is no assumption in edgeR that the sample sizes are equal and, in this case, both sample sizes are substantial. The main factor that limits the power of a DE analysis is the smaller of the two sample sizes rather than the ratio of the two sample sizes.

We have used edgeR even for n=1 vs n=2, although that is obviously an extreme scenario that I only do when I have to.