Bioinformatics Approach To Summarize Functional Diversity Of A Gene

Question

[Khader Shameer]

I am looking at a particular gene (n=1), the gene is involved in a variety of molecular functions like signal transduction, molecular binding etc. What will be the best way to summarize the function of such a gene ? I am thinking of analyzing individual interactome of it's protein product(less direct interaction, but good number of interactions via STRING (confidence score=0.4)) and use BiNGO to derive the over-all function. Do we have any better approach to illustrate functional diversity of a single gene ?

Answer 1

[Lars Juhl Jensen]

The approach you suggest is sound, however, I personally do not think that a list of GO terms is a particularly good way to summarize how a gene works. You already write yourself that the gene is involved in a variety of molecular functions, which I assume are GO molecular function terms. Including GO terms from its interaction parters will probably only give you an even longer list.

The best approach depends on whom you need to summarize it for. If it is for a biologist, and the gene is involved in signal transduction, the best way is probably a schematic drawing of how it is involved in which signaling processes (assuming that this is known). I know that this is not a nice computer-readable, ontology-based way of storing information, but when it comes to communicating complex information to a human being, it is difficult to beat a good figure.

[Chris Evelo]

Lars might still be right though. 30 Molecular functions are not necessarily 30 individual interactions. Some interactions may have been classified as multiple GO functions. You could draw a Pathway using PathVisio with your gene in the center. In that way you could also illustrate the processes a bit more. It would be intuitive to read for biologists and they could click through to the packpages to understand the biology better. You could use the automated suggestion plugin to get data from e.g. IntAct and (maybe extend it with STRING and Bingo). And... It would in fact be computer readable.

[Khader Shameer]

Thanks for your thoughts Lars. A single figure may not help me here, because there are more than 20 GO terms associated with the genes. True that BiNGO approach will further expand the list of GO terms, I am trying now to summarize the function using GO structure figure from BiNGO - I find it quite intuitive to explain the enriched nodes as prominent functions.

Answer 2

[Qdjm]

You could also try GeneMANIA.

It's like STRING, but has more data and it does the GO enrichment analysis for you (go to the functions tab). By default, it doesn't incorporate information about protein domains but you can change that by activating that network in the advanced options panel.

A GeneMANIA single gene lookup for BRCA1.

[Khader Shameer]

GeneMANIA looks interesting. Thanks for the link.