I’m attending the BioHackathon 2010, the topic is “Semantic Web”.
Everthing is fantastic, yeah you need a dozen of servers to crunch your data but that’s not the real problem. The real problem is how can we use that stuff ?
There are different answers to this simple question.
You can build a fancy/functionalÂ interface like BioGateway Semantic Systems Biology by Erick Antezana . Actually this is a great work that puts in the hands of “scientists” the power of Semantic Web. The facade is just a pice of the huge work done by Eric.
Then there are the people in the middle, blue helments of the OpenBio* projects like BioRuby, BioPython, BioPerl, BioJava that need to come up with a way to use the so called Semantic Web.
I’m a BioRuby and I was in charge to consume this SemWeb. Googling around I found ActiveRDF a very cute library which solved part of my/our problems. After some days of panic and bugs now BioRuby can consume in an Agile way every SPARQL endpoints. Thanks Eyal Oren, Renaud Delbru, Sebastian Gerke, and Benjamin Heitmann for your great work.
Query Bio2RDF GeneId endpoint.
I know, the cose is messy but it’s just a test…
Hey, but where is the agile stuff? Right there
myGeneSemWeb is an array of triplettes. Just grab the first one:
Not yet agile, yessss I know… but now?
At this point you can browse your object dinamically, the only limitation is moving from one domain to the other. We are using the geneid endpoint if you want to move to kegg it’s not yet possiblem but is something we want to do.
There are a lot of things to do:
* design the integration with bioruby
* handle multiple endpoints
* test federation and create federated queries
* provide to the end user a friendly api.
* We all are very happy to have met Brad Chapman and Peter Cock from BioPython Dev Team, we plan to develop the same/similar API and work together. You can find Brad’s work here
* I’d like to implement this by default as autocompletion into irb
* Map RDFS classes to BioRuby objects.