Combining GO and KEGG enrichments

Question

[David R]

Hi,

I have a set of DEGs for which I have carried out GO and Kegg enrichment analysis. Thus, I have twos sets of enriched terms for those genes, one with GOs and another one with KEGGs. I would like to produce a network in which I could put enriched GOs and KEGGs as nodes. Two nodes would be then connected when at least one gene is affected to both nodes. I would be specially interested in edges connecting GO with KEGG nodes. I could also use the number of genes in common between a GO and a KEGG as edge attribute (colour, thickness...)

Do you think this kind of network is feasible? I am not sure what the columns of my table(s) should be.

Thanks!

David R.

[Yogi]

Seems a bit strange to me but I can kinda see it.

Wouldn't trust the idea of just a single gene being part of two pathways giving you edges.

That would be a massive network probably.

You'll need to create a table specifically based on the format that a downstream network creation package will work with.

I definitely think it's important to use edge strengths.

It may even make sense to specifically visualize the node connections that have extreme overlap.

[David R]

Thanks a lot Yogi. I understand you are thinking about "network" package in R, for instance. I'll give it a go. Thanks for the hint. Best, David

Answer 1

[xanderpico]

Sorry for the late reply, but if you are working with these data in R, then I'd recommend the RCy3 package (in Bioconductor) to help turn any dataframe into a Cytoscape network. In Cytoscape, you can easily model the shared nodes among GO terms and pathways as weighted edges and then have access to a ton of visualization and layout options to make compelling figures. All of these options are accessible from R via RCy3.

[David R]

Thanks a lot. Yes, I have started using RCy3 and it seems great. It will also improve Cytoscape traceability on my work .