Hello,

I am interested in searching for all GO terms associated with a specific cell type or tissue.

In

http://wiki.geneontology.org/index.php/Annotation_Extension:_Capturing_cell_and_tissue_types

I see that in the "Extension" field of GO I would find this type of information, referenced to other ontologies as well.

However, I tried to obtain the same information with GO.db and rols, with no success. Specifically, I tried to replicate the use cases in the above link. 

I tried to retrieve all the terms from GO.db and then checked all the GOXXPARENTS, to see if any of the parents has the "occurs_in" relationship with its children, and I found none.

I feel that I am completely off, does anybody have any suggestion on how to retrieve the annotation extension information from R?

Thanks,

Mic

Hi Michele,

The traditional GO doesn't have any ontology for tissue type.  There is an ontology for biological process (BP), molecular function (MF) and for cellular component (CC).  But there is not a GO for tissue type or anatomical location.  It looks like right now GO is essentially telling people to use other peoples terms for annotating those kinds of things when describing genes (IOW using other peoples IDs and ontologies such as Uberon or cell ontology).

Since that is the case, you can't directly look up GO terms that are associated with a specific tissue type unless the GO actually added a new ontology, or you had some domain specific knowledge to leverage (which could include annotations from other ontologies).  But the gene ontology alone cannot indicate tissue.  So if for example, you wanted to study GO terms that relate only to muscle cell development, you have to know somehow which specific processes are specifically involved in that and then go after those terms (and their child terms).

As for looking up all the specific children for a GO Term here is how you could do that:

library(GO.db) 
muscleProcessID = "GO:0003012"
muscleIDs = unlist(mget(muscleProcessID,GOBPOFFSPRING),use.names=FALSE) 
muscleIDs = c(muscleProcessID,muscleIDs)

And once you have those, you could then take those IDs to something like an org package and look up things like associated gene IDs etc. like this

library(org.Hs.eg.db) 
select(org.Hs.eg.db, muscleIDs, 'SYMBOL', 'GO')

Hope this helps.   If you would like to contribute some more packages to help us also cover some of these other ontologies, then please feel free to contact me. 

Dear Michele,

I am afraid that rols will not be of any help here. The ols webservice does not provide access to the term hierarchy. You can get cross references (termXrefs), but I don't think this is will get you want you want.

Have you come across any software or online ressource that implemented extensions across ontologies?

Hi Laurent,

thanks for your reply.

I found that AmiGO supports such information: if you look at the

http://amigo.geneontology.org/amigo/gene_product/UniProtKB:P63027

page, downloading the table of results shows that some of the terms associated with the gene product are cross-referenced to other ontologies (UBERON and CL).

However, I was not able to retrieve the same information from GO.db.

I was thinking of using SPARQL to find this information, but I admit that I am not proficient enough in SPARQL queries.

Any thought?