Although we didn't have time to implement this example before the
C-SHALS meeting, as we discussed at our BioRDF call today, we will
implement this example and include it in the BMC Bioinformatics paper
(on query federation) that we're working on. Eric has updated his
FeDeRate code so that performance has been improved significantly.
The receptor tables (in csv format) are listed at:
http://esw.w3.org/topic/HCLSIG_BioRDF_Subgroup/QueryFederation/iuphar_receptors
Eric, please use the "Class C" table and load it in your relational
database. Also, Matthias and Rob will help in this implementation effort.
Cheers,
-Kei
Matthias Samwald wrote:
(Moving this discussion to the public mailing list)
To follow up the teleconference we just had...
Here is an example of finding links between DBpedia receptors and the
IUPHAR receptor table:
* Look at http://dbpedia.org/resource/GABAB_receptor
* Look at property http://dbpedia.org/ontology/uniprot it has the
values "O75899" and "Q9UBS5"
* Now look at the IUPHAR "Class C" spreadsheet
* You will find these exact identifiers in the "Human nucleotide" and
"Human SwissProt" columns of the matching IUPHAR entry, respectively
(that should be columns 8 and 10, if I counted right).
-> Mapping accomplished!
Note that only some of the receptors mentioned in
http://neuroweb.med.yale.edu/res/senselab_receptor_to_DBpedia_mapping.owl
can be mapped to IUPHAR with this procedure (not all receptors have
sequence entries, because they are multimeric complexes). But for
those entities where it works, we can create a federated query that
spans NeuronDB (OWL file), DBpedia (SPARQL endpoint), IUPHAR (direct
query of relational tables / CSV), PDSP KiDB (OWL generated from CSV)
and others -- quite useful and a good show case for query federation.
Hope that helps,
Matthias Samwald
DERI Galway, Ireland
http://deri.ie/
Konrad Lorenz Institute for Evolution & Cognition Research, Austria
http://kli.ac.at/