Functional annotation of pathways in a genelist

Question

[tejas_j]

Hi everyone!

I'm currently doing GSEA using genelists from msigdb. One of the issues that I'm having is that the enrichment analysis gives a large number of pathways, some of which are expected for the data I have. Is there a way to annotate these pathways based on their function such as structural, immune, etc. so that I can filter out the obvious pathways and get a smaller list.

Thanks!

Answer 1

[hibernicah]

Gene ontologies are organised into a hierarchical tree structure, which means that smaller gene sets are usually a subset of a larger, broader set that may be helpful to categorise your results. You can also explore annotated relationship between them, see: http://geneontology.org/docs/ontology-relations/

The C5 MSigDB set is composed of the gene ontologies terms. For example: GO_3_UTR_MEDIATED_MRNA_STABILIZATION corresponds to GO:0070935. If you want to avoid browsing ontologies of interest one by one, you use for example Quick GO API or GO.db Bioconductor package.

[tejas_j]

I'm sorry if I wasn't clear. I currently use the hallmark and C2 sets and get results for enrichment. What I want to do is filter the resultant pathways based on their function. For example, look at the pathways that are involved in stress response. In other words, if I have a GMT file with a list of pathways and their genes, how do I mark these pathways based on what they do so that I can do further analysis.

Thanks!

[hibernicah]

What do you mean by further analysis? Do you see eg. stress response associated signatures enrichment and you want to summarise them for visualisation or do you want to draw more general conclusions from your data that you were unable to see so far?

[tejas_j]

As of now I get 100+ pathways which are significantly enriched. I would like to categorize and summarize these pathways so that I can see what they do functionally. For example, in my data stress response is expected and I'm interested in pathways which are not related to stress response. So if I had a way to categorize the pathways, I could simply filter out stress response and then have a better look at other pathways. Thanks!

[hibernicah]

In order to narrow down a list of GO/pathways you can explore the pathways in terms of GO hierarchy as I mentioned earlier or simply adding more stringent threshold for significance, minimum or maximum number of genes in the pathway that was in an input list etc.