W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > June 2006

[BIONT] [BioRDF] a model for integrating DB's

From: John Barkley <jbarkley@nist.gov>
Date: Thu, 1 Jun 2006 09:58:18 -0400
Message-ID: <0eb401c68583$6fac7160$8a3a0681@ncsl.nist.gov>
To: <public-semweb-lifesci@w3.org>

>From this week's BIONT teleconference discussion, I put together an example
illustrating one way RDF/OWL could be used to integrate databases (DB) using
vocabularies. This approach is inspired by the merged/neuron/cocodat
knowledge bases (KB) at http://128.36.123.50/NeuroWeb/.
merged.owl integrates neuron.owl and cocodat.owl using the OWL construct
"sameAs" to relate similar terms in neuron and cocodat.

This example illustrates a similar approach to integrating OMIM and Brain
Atlas using UMLS disease terminology. Please note that, unlike the NeuroWeb
KBs which relate terms between neuron.owl and cocadat.owl,  this example
relates terms in the DBs being integrated to a "standard" vocabulary. The
URL is:

http://hissa.nist.gov/jb/neuro-db-integration-model/m1-integrator.owl

Characteristics of this example include:

1. The integrator relates UMLS disease names with OMIM and Brain Atlas
disease names. SPARQL can be used to find your way around, e.g.:

a. Find all Brain Atlas and OMIM names and their HTML URLs for the UMLS name
"Alzheimer_Disease":

SELECT ?d ?u
WHERE { umls:Alzheimer_Disease owl:sameAs ?d .
OPTIONAL {?d omim:has_OMIM_entry ?u .}
OPTIONAL {?d ba:has_URI ?u .}
}

Result:

ba:Alzheimer_disease
http://www.med.harvard.edu/AANLIB/cases/case29/mr1-tc1/017.html
ba:Alzheimer_disease
http://www.med.harvard.edu/AANLIB/cases/case40/mr1-tc1/033.html
omim:_607822   http://www.ncbi.nlm.nih.gov/entrez/dispomim.cgi?id=607822
omim:_104300   http://www.ncbi.nlm.nih.gov/entrez/dispomim.cgi?id=104300

b. Find all Brain Atlas and OMIM names for the OMIM name "_104300" and match
with UMLS names:

SELECT ?n ?n1
WHERE { omim:_104300 owl:sameAs ?n .
               ?n owl:sameAs ?n1 .
}

Result:

umls:Alzheimer_Disease   ba:Alzheimer_disease
umls:Alzheimer_Disease   omim:_607822
umls:Alzheimer_Disease   omim:_104300

(To keep the examples short, the PREFIXs have not been included.)

2. Only the integrator need "know" about the OMIM, Brain Atlas, and
vocabulary KBs. None of these need reference each other. It may be the case
that a database's KB may already exist and not have to be created by us.

Protege 3.2 beta with the SPARQL panel was used to create this example and
its associated queries.

jb
Received on Thursday, 1 June 2006 14:01:24 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:00:43 GMT