Standard methods for filtering enriched GO terms based on similarity scores?

Question

[mbrando17]

I have a list of enriched GO terms computed using any gene enrichment tool. I can compute similarity scores for my list of enriched terms using GOSim. And then based on the GOSim similarity scores, I would like to filter my list of enriched terms to remove redundant terms and reduce to a smaller, more meaningful sublist. Are there any precedents or standards for doing this? I can't seem to find anything.

Answer 1

[Renesh]

You can use Q-value for filtering enriched GO-terms.

[mbrando17]

I'm assuming my list of enriched GO terms already consists of those enriched terms with adj. p-value (or other statistic) < threshold. I would like to further filter (or essentially re-rank) the list using term similarity scores so that if a set of terms is very similar, one term is kept and the others are removed (or given a lower ranking). I'd like to avoid making arbitrary choices if possible, so I'm just wondering if anyone has seen this done before.

Answer 2

[Woa]

See if ReviGo can solve your problem:
http://revigo.irb.hr/

[mbrando17]

That's just what I'm looking for. Thanks!