Ontology Of Biological/Biomedical Identifiers

Question

[Joachim]

Hello,

I am currently extending my web-service and online resource http://www.opacmo.org/ to support downloads in RDF n-triple format. Most of the data in opacmo already comes from ontologies (GO, DO, ChEBI), but there are also PubMed Central identifiers, Entrez genes and species ids from NCBI's taxonomy present. For the latter three I am not quite sure how I would express their "type" in the RDF output.

For example, if I have an Entrez gene ID, I could give it an RDF type via "is_a" as "Entrez Gene" from the NCI Thesaurus (first search result of http://bioportal.bioontology.org/search?query=entrez+gene&commit=Search). However, if I look for "Pubmed Central" in the BioPortal, then I only get "pmc" from the EDAM ontology as somewhat meaningful result, but EDAM's "pmc" refers actually to the article format and not the identifier.

In a sense I am really looking for an ontology that covers most common biologically/biomedically relevant identifiers. EDAM is already a good start, because it can identify Ensembl gene IDs and PubMed IDs -- even though it does not help me very much with my problem.

Does anyone know if there is a single ontology (like EDAM) that I could use to semantically annotate PubMed Central identifiers, Entrez gene IDs and species IDs from NCBI's taxonomy? Is there an alternative way of giving these identifiers a type in RDF?

Thanks, Joachim

Answer 1

[B. Arman Aksoy]

BioPAX utilizes MIRIAM in order to accomplish the standardization of external references. What the guidelines commonly suggest is to use MIRIAM URIs as RDF IDs to the objects that represent external references. I think it will be much better for you to see it in action so here is a simple BioPAX export sample:

http://www.pathwaycommons.org/pc2/get?uri=urn:miriam:uniprot:Q06609

...
<bp:PublicationXref rdf:about="urn:miriam:pubmed:11158580">
 <bp:id rdf:datatype = "xsd:string">11158580</bp:id>
 <bp:db rdf:datatype = "xsd:string">PubMed</bp:db>
</bp:PublicationXref>
...
<bp:ProteinReference rdf:about="urn:miriam:uniprot:Q12841">
...
 <bp:organism rdf:resource="urn:miriam:taxonomy:9606" />
</bp:ProteinReference>
...

I, as a humble bioinformatician, find this approach quite successful and intuitive.

Hope it helps.

[Joachim]

Thank you for pointing out BioPAX. I had a look at their ontology (http://www.biopax.org/release/biopax-level3.owl) and your example and it is indeed a very intuitive approach to the problem.

Answer 2

[Pierre Lindenbaum]

You could get your inspiration from the ontologies used by Bio2Rdf

Example for NCBI GENE: http://bio2rdf.org/rdfxml/geneid:15275

<rdf:Description rdf:about="&lt;a href=" http:="" bio2rdf.org="" geneid:15275"="" rel="nofollow">http://bio2rdf.org/geneid:15275">
(...)
<describes_gene_or_gene_product_mentioned_by xmlns="&lt;a href=" http:="" purl.org="" science="" owl="" sciencecommons="" "="" rel="nofollow">http://purl.org/science/owl/sciencecommons/" rdf:resource="http://bio2rdf.org/pubmed:17626619"/>
(...)

[Joachim]

Thanks, Pierre. Bio2RDF looks interesting, but I find the RDF output less accessible compared to BioPAX. However, I might get in touch with Michel again, especially as they seem to reboot Bio2RDF now (http://bio2rdf.blogspot.com/2011/09/week-we-held-our-first-virtual-meeting.html).