I understand now better the problem. I don't have a set of control genes, I am following the procedure of finding "in-silico empirical" negative controls, but in the first-pass DE analysis I am not finding any DE microRNA. See:
I have another question Davide, maybe too naive, but I need it just to confirm. I am a bit confused with the direction of the contrast while using the tool, I am reading edgeR user guide and I am not fully sure, when you are specifying coef=2 in your example (let<-glmLRT(fit, coef=2)) you are doing Treatment against Control, right? Meaning that if logFC is negative you will have a increased gene expression in treatment and if it is positive the other way around...?