I'm struggling to make sense of some GO-term enrichments results I have. For the disease (an inflammatory disease) I'm studying, I get GO terms which don't really seem relevant to it. For example, I have a lot of cancer terms and I'm thinking these may reflect increased proliferation, but how can I justify this in a paper? I also have a lot of terms related to specific infections (eg. HPV and CMV) which may reflect inflammation, but I run into the same problem - how can this be justified?
I have tried doing literature searches for the terms themselves, but the papers are very specific to the disease studied and I am unsure how to apply it to my disease of interest.
This, unfortunately, is almost always the case with GO analysis. If you look into those cancer related terms, I suspect you'll find that they have lots of inflamitory genes in them.
Once upon a time, I was analysing data from a cancer study, and ran a go analysis, and several of the terms that came back as cardiac related. When I looked into which genes within those terms were DE, it was the integrins. As cancer cells loose contact inhibition, Integrins are important cancer genes, but it turns out they are also important for various connective tissue disorders that affect the heart. By clustering together terms on the basis of the DE genes in them, I could show that all the DE genes in the heart terms were also found in cancer terms.
Even so, this is all building just-so stories. GO analysis is useful for building hypotheses, or for confirming at a results list is enriched in vaguely the right things, but it will rare provide a solid finding in its own right.
