On 15 February 2010 11:56, Arek Kasprzyk <[email protected]> wrote:
> On 14/02/10 5:58 PM, "Peter Ansell" <[email protected]> wrote:
>
>> Hi all,
>>
>> I am interested in creating an RDF output format to BioMart but I am a
>> little lost looking through the source code about where to start. To
>> start off with I would like to add functionality to support URI's for
>> columns to match RDF predicates, and add URI support for identifiers
>> in the database so that they can be identified as items in the RDF. I
>> know it will require some changes to the XML definitions, and an
>> output class to suit, but I am not sure what needs to be done other
>> than that, or how much I have to rely on the current code for doing
>> that.
>>
>> Thanks,
>>
>> Peter
>
> Hi Peter,
> We are in the middle of re-writing the configuration system so we can easily
> extend it contain the predicate information. I was thinking about your
> problem recently and thought that rather dumping marts to RDF would probably
> be better to think about either SPARQL end point or some other web services
> based equivalent to make mart look semantic web compliant. For instance we
> could extend the existing web service to return RDF triples. Would that
> help?

After the configuration has support for predicates and mapping item
identifiers to URI's, the SPARQL endpoint would be the next logical
step, because from that level you can generate RDF documents as you
like anyway. Let me know if you want any comments on the new
configuration code.

The Filters mechanism in Biomart seems to map conceptually onto SPARQL
Filters so the query performance of the SPARQL endpoint could be
similar to current filtered queries.

Given the number of bioinformatics databases that host their data
using Biomart it would be a very useful from my point of view to have
SPARQL/RDF support.

Cheers,

Peter

Reply via email to