Reactome RDF

[Alan Ruttenberg]

Available at http://sw.neurocommons.org:8894/sparql 

I'll install it the neurocommons once we've caught any further errors. This is the basic info, for reference.

Couple of queries to get you going:

Default-graph: http://purl.org/commons/record/reactome/ (should have named it better)

prefix re: <http://purl.org/commons/ontology/reactome/>

prefix rei: <http://purl.org/commons/record/reactome/>

prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> 

prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> 

SELECT distinct ?class WHERE {?s rdf:type ?class}

SELECT distinct ?item WHERE {?item rdf:type re:DataModel}

SELECT distinct ?label WHERE {?item rdf:type re:RegulationType. ?item rdfs:label ?label}

SELECT distinct ?p WHERE {?s ?p ?o}

SELECT ?ontology  WHERE {?s re:ontology ?ontology}   # unicode encoded version of the ontology as rendered in GKB

Loading it in to virtuoso:

DB.DBA.TTLP_MT(file_to_string_output('/work/reactome/reactome.nt'),'http://purl.org/commons/record/reactome/' ,'http://purl.org/commons/record/reactome/',0);

Creating the rdf using d2rq:

sudo ./dump-rdf -f N-TRIPLE -o reactome.nt  -m file:///work/reactome/d2r-server-0.4/reactomemapping.n3 -b http://purl.org/commons/record/reactome/

(note the N3 output, while prettier, exercises a bug in d2rq)

Attached: reactomemapping.pl (scripts a rewrite of the default d2rq mapping)

reactomemapping.n3(the output of the script, used as input to d2rq)

[Alan Ruttenberg]

Forgot to mention - wouldn't have happened were not for Andrea's help - Thanks!

-Alan

<reactomemapping.pl>

<reactomemapping.n3>

[Alan Ruttenberg]

Couple of minor corrections to the d2rq mapping - will regenerate and reload the endpoint probably tomorrow.

I've also generated an OWL specification of the database record structure based on the protege frames description that was embedded in the query result for SELECT ?ontology  WHERE {?s re:ontology ?ontology}

http://svn.neurocommons.org/svn/trunk/convert/reactome/reactome-records.owl

The mappings and code are in the same directory.

Let me know if (when:) you find any mistakes.

With this in place, we should now be able to start doing sanity checks and conversions to a more oboish representation.

-Alan

On Mar 15, 2008, at 4:46 PM, Alan Ruttenberg wrote:

<reactomemapping.pl>

<reactomemapping.n3>

[Alan Ruttenberg]

On Mar 16, 2008, at 3:43 AM, Alan Ruttenberg wrote:

> will regenerate and reload the endpoint probably tomorrow

Ah yes, no time like the present. Also loaded the owl file.

Note: Default graph name is now: http://purl.org/science/graph/
reactome-records

Here's one that gives a bit of documentation on some of the reactome
record classes

prefix owl: <http://www.w3.org/2002/07/owl#>

prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>

SELECT distinct ?class ?comment WHERE {?s rdf:type owl:Class. ?class
rdfs:comment ?comment}

http://tinyurl.com/3cr66p will get you the results in XML

-Alan