Hi,
I have some differential exon usage results generated by DexSeq, but when I try to make sense of the them using the plotDEXSeq function, it becomes puzzling to me why some exon fragments are significant and others are not (currently setting FDR < 0.001, because my data has way too many significant hits, which is another part of my confusion). For example, here is what a gene looks like (posting img for the first time and hopefully it will look fine). I'm not sure why E002 is not significant while E003 is, as they look similar to me in terms of distribution of normalized counts and E002 actually has more different exon usage estimate between the two groups. I guess that has to do with dispersions estimates for these two fragments? I should also note that there are 3 samples in the red group and 13 samples in the blue group, but I thought unbalanced group sizes only reduce sensitivity in general?
Thank you so much for your help!!