Choosing the right pathways among gene set enrichment analysis results

Question

[jabbari.parnian]

Hello everyone,

I have performed gene set enrichment analysis on my RNA-seq data using the GSEA software by Broad Institute. Depending on the comparisons and gene set database we choose, there is an overwhelming number of pathways returned. I was wondering if there is a way to choose among these pathways to do further experiment such as pathway inhibition, etc? Any suggestions are appreciated.

Thanks

Answer 1

[Jean-Karim Heriche]

Statistical significance doesn't mean biological relevance so one alternative approach is to rank the pathways by score and look at whether there's an obvious cut-off such as an elbow or big gap and focus on those pathways above this cut-off. Also you can quite arbitrarily decide to only consider the top n pathways because this is all that your resources will allow you to explore.
Also consider looking at Gene Ontology terms enrichment. While working with pathways as gene sets is easy, the downside is that you have no idea about the relationships between pathways (and these differ between different resources). For example, your top 10 pathways could be involved in the same two biological processes without this being obvious from the way they are named. GO will give you this relationship though the trick is to use an enrichment approach that adequately handles the graph structure of the ontology to focus only on the most relevant terms.
Other than that, without using additional information, gene enrichment analysis from expression data is often simply a fishing expedition. Unless you're doing RNA-seq just for the sake of it, you probably have some additional information that is relevant to the biology or question of interest so make use of it.

Answer 2

[i.sudbery]

I hate to say it, but I'm not sure there is a really good solution to this, other than knowing the biology of the system/disease/response well enough know what might be important.

Answer 3

[jared.andrews07]

There are also ways to collapse similar pathways, like the collapsePathways() function from fgsea.