Entering edit mode
3.4 years ago
Francois Piumi
▴
70
Hi, I'd like to associate a "general function" to a gene. I can find the GO terms associated to one gene, and then Revigo allows me to keep the most general term ("frequency"). But it has to be done gene-by-gene and it's very time consuming (you can't work with gene lists). Do you have any idea to process this with lists of genes? Thanks
I'm sorry but that's absolutely not what I asked. I don't want an enrichment analysis, there's already tons of software to do this.
I want a final table giving one function for one gene. I have a list of genes and I want the best revigo score (for example "frequency", "uniqueness" or "dispensability") for each gene.
edit: Can you not download the reduced go terms from the REVIGO web-tool and map them to genes after?
Am I maybe not understanding what REVIGO does? It reduces GO terms, and this R package is a software package that does that also. The authors of the package say that it mimics REVIGO.
I am going to work with the package to see if I can extract these "frequency", "uniqueness" or "dispensability" terms from their internal data structures. If they mimic core functionality of REVIGO, then maybe they will be present.
Further, this package does not conduct standard GO enrichment analysis, but reduces GO terms.
Thanks @drew.b.ferrell for your help. I have 258 genes to submit to REVIGO and I don't want to submit them one by one. Output of rrvgo seemed an enrichment analysis to me, not a gene functional annotation.
I cannot find those terms in this package, at least. I think your best bet is to map genes from the REVIGO web-tool output.
I'm confused why you mention working with gene lists, as this tool uses GO term lists. Could you clarify this bit?
Assuming you have GO term lists, a gene to GO mapping file, and a little programming experience, you can map these terms ("frequency", "uniqueness" or "dispensability") to genes.
Sorry I forgot to mention that first, I got all the GO terms corresponding to one gene (I use Biomart or the Genecards website). Then I enter all the GO terms for one gene in REVIGO. I have to do this for each gene, one by one. Then I keep only the best "frequency" or "uniqueness" score.
Francois Piumi : Please use
ADD REPLY/ADD COMMENTwhen responding to existing posts to keep threads logically organized.Do you have p-values? You may be able to label each GO ontology set with some ID, using the p-value column as an ID instead.
Edit: they don't return the input p-value per GO term for some reason, so this wouldn't work. What you need is web-scraping, so you can access their web-tool from some programming language and automate each gene list. I am not sure how to do this.
No I don't have any p-value per GO term, just the name of the GO term. I remember that I already contacted Rajko Horvat from REVIGO and he answered very quickly. I will write him again to know how we may arrange that.
Right, sounds good. I suggest also being open to other R packages that conduct this analysis (GO term semantic similarity), as it seems this particular tool cannot be automated.
Thanks Drew!