Search Mailing List Archives

Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[bioontology-support] Questions about SPARQL endpoint

manuelso manuelso at
Wed May 23 16:35:40 PDT 2012

Hi Josh,

Thanks for the feedback. See inline answers below:

On May 23, 2012, at 3:52 PM, Josh Mandel wrote:

> First off, congratulations! is shaping up to
> be a really great resource. I love the built-in examples, slick UI,
> and (most important) substantial content!
> A few of questions / issues:
> 0.  Having the UMLS in one place makes for some very convenient
> queries!  For example, I'm using as
> an example to figure out whether a medication list contains any
> statins.   Hierarchical queries (e.g. to sort a problem list by body
> system) would get much easier with support for either subclass
> inferences *or* SPARQL 1.1 property paths (?a rdfs:subClassOf* ?b.)

Good example. We are working on providing transitive reasoning for subClassOf and other similar predicates. We will have this ready very soon, in few weeks. It will not be in the form of property paths but it will be standard SPARQL.

> 1.  The documentation (and examples) don't make it clear what the
> default graph is for queries.  It looks like the default graph is the
> union of all named graphs; is that right?  Are any triples asserted
> directly into the default graph?

The default graph is the union of all graphs in the RDF store. There is no such thing as one unique default graph in our RDF store. I'll update the documentation to reflect this. Thanks.

> 2.  It's my understanding that
> generates a strict subset of the triples that
> hosts.  Is this right?  

Right, the UMLS2RDF project takes care of only UMLS ontologies. In the RDF store there are also OWL and OBO ontologies. Basically all the ontologies that you see here:

> Is there any
> way to generate or obtain the entire set?

Yes there is one. You can download all the ontologies from BioPortal transform them into triples and create an equivalent database to This is essentially what we do in order to generate the RDF store. There is one REST service in BioPortal that allows you to download the ontology data that was uploaded to BioPortal:

Moreover, there is another service that allows you to download the RDF file that represents one ontology in the triple store:

Take into account that, due to licensing issues, not all UMLS ontologies are downloadable.

> 3.  The sparql endpoint ignores the `output` GET variable (but pays
> attention to the Accept: header).

Right, the content negotiation is handled by HTTP headers. "Output" var in GET calls is ignored.
Maybe we should be able to handle both mechanisms ... we will give it a thought.

> 4.  The sparql endpoint fails to set Access-Control-Allow-Origin: *
> when there is an error (missing apikey; typo in query, etc).  This
> make debugging a browser-based app difficult (because the error
> message doesn't make it to the client code; the request simply
> aborts).

Point taken. I've issued a bug for this and we will get it fixed.

> 5.  Because of (3) (and also because it appears to time out against
> alphasparql.bioontology.otg) the JS sparql client fails:
> I've submitted a pull request with a simplified (working) example:

Good stuff. Pull accepted. Thanks a lot!

> 6.  github SPARQL examples still point to alphasparql

I thought I had changed all of them but it looks I missed some.



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the bioontology-support mailing list