Public Sparql End Points For Pubmed?

Question

[Egon Willighagen]

Bio2RDF (doi:10.1016/j.jbi.2008.03.004) has a SPARQL end point for PubMed, but the amount of data is so large, that a query for all properties already gets rejected because it would take too much time.

Are there alternative SPARQL end points publicly available?

Answer 1

[Pierre Lindenbaum]

Not tested:

http://linkedlifedata.com/ claims that you can run a SPARQL query on their pubmed graph.

Search and explore over 5 billion RDF statements from various sources including UniProt, PubMed, EntrezGene and 20 more...

Perform complex SPARQL queries and retrieve more than one billion RDF resources.

[Egon Willighagen]

I just tested this, and it works fine! Thanx for the tip!

Answer 2

[Joachim]

This is not a ready solution and perhaps not what you are looking for, but nevertheless:

1. Have a look at www.pubmed2ensembl.org, which links publications to genes and vice versa
2. Get BioMart 0.8rc7 at www.biomart.org and install it locally
3. Use MartConfigurator to import the pubmed2ensembl datasources as URI sources*
4. Use BioMart 0.8rc7's SPARQL endpoint to access pubmed2ensembl

The data stays on www.pubmed2ensembl.org. SPARQL queries to your local 0.8rc7 BioMart will automatically fetch the data from the remote source.

BioMart's SPARQL support is currently fairly limited, because it would otherwise not be possible to support SPARQL endpoints on the larger databases (s.a. Ensembl) on central.biomart.org or dcc.icgc.org plus we need to make sure that federation works too. Essentially, you can only query leaves of the ontology (Literals), but you cannot traverse the graph using SPARQL. Have a look at the auto-generated SPARQL queries and you will see what I mean (http://www.biomart.org/rc6_documentation.pdf -- rc6 docs == rc7 docs).

While this is not directly querying PubMed, pubmed2ensembl would at least give you a view on MEDLINE + PMC as seen through "gene glasses".

Joachim

*There is one quirk. If you are going to use this solution, then I can tell you how to enable importing the partitioned species data as it appears in pubmed2ensembl. There is one feature that you need to activate in the source-code of BioMart in order to seamlessly integrate all species-info from a 0.7 mart into a 0.8 mart, which is off by default because we had no time to test it for the release.

[Egon Willighagen]

Yeah, I heard about SPARQL support for BioMart! I think that's a smart (and important) move! Looking forward to a finished product.

[paulr]

Just wondering what the status of the SPARQL support is?