Using GOseq on only a sub-fraction of edgeR DE results

[sindrle]

Hi!
I got a question from one of my lab mates if I could take out a certain type of genes (all coding for one certain type of proteins) and run GO. This is a very broad category of genes, about 50% of the total DE genes was this type.


This is what I thought:

1. Find DE genes that codes for these proteins
2. Set ALL OTHER gene to NOT DE, even if they are.
3. Run GOseq


Is there some obvious problems I will run into?

thanks