RE: [rdf] Re: proposal for standard NCBI database URI

Hi All,
 
I'm new to this list and I only did a cursory reading of the archives so I 
apologize if I add anything that has already been addressed.  As some of 
you know, Stephen Evanchik and myself are the lead developers on the LSID 
Toolkit.  I have implemented a prototype LSID Resolution Service for 4 
NCBI databases (Pubmed, OMIM, Protein, Nucleotide) and a fifth (Gene) is 
in the works and should be available shortly. Each of these returns RDF as 
metadata if the LSID if dereferenced.  As with all LSID implementations, 
the lion's share of the work has been coming up with ontologies for the 
metadata.  I did a time-consuming but naive first approach by 
transliterating the various XML formats into OWL.  You can find these 
ontologies at 

http://lsid.biopathways.org/ncbi/ontologies/

NCBI has recently upgraded their web service from 1.3 to 1.4 which is why 
the services have been broken as of late.  I've upgraded the code, but I'm 
waiting for them to address a bug I'm having with the OMIM and Gene Web 
Services.  In the meantime, I would appreciate any feedback on the 
ontologies.  Immediate modifications that I have considered include, 
condensing the predicates and using DC and RDFS predicates where possible. 
 I'm sure that there are other ontologies out there that I should also 
make use of but I don't know what those should be so any advice would be 
appreciated.  These LSID resolvers should all be available soon at a new 
home for lsid.biopathways.org that has much better bandwidth.

A collegue of mine is also interested in the outcome of  discussions 
regarding these ontologies as he is working on a SPARQL engine that we 
could use to make federated queries against NCBI, also via Biopathways.


  Thanks,
 Ben Szekely

PS, Obviously, as developers of the LSID toolkit, we would favor LSIDs for 
NCBI URIs for any number of reasons!