Open Bio* Semantic Web tools

Participants

(Please add your name!)

• Toshiaki Katayama
• Erick Antezana
• Shuichi Kawashima
• Mitsuteru Nakao
• Brad Chapman
• Peter Cock
• Jan Aearts
• Kyung-Hoon Kwon
• Raoul Bonnal
• Christian Zmasek
• Keun-Joon Park

RDF libraries in each language

(Please add other languages as well!)

Name of the available libraries:

• Ruby
  • ActiveRDF: may depends on Rails
  • We have a lot more. see -->  http://raa.ruby-lang.org/search.rhtml?search=rdf
• Python
  • RDFLib  http://www.rdflib.net/
  • SPARQLWrapper  http://sparql-wrapper.sourceforge.net/
• Perl
  • RDF: generic RDF library
  •  ONTO-PERL: it might be used (it is, however, OBO-centric  http://bioinformatics.oxfordjournals.org/cgi/content/abstract/btn042)

Functionality

• RDF reader (in)
• RDF generator (out)

language	in	out
Ruby	?	?
Python	o	?
Perl	o	?

Tools to manipulate RDF data

• RDF grep
• RDF diff
• RDF sort
• RDF uniq
• RDF wc (triple count)
• RDF cat (combine)
  • (most of) these commands will output RDF graph :)
  • "sort" before "diff" is better.
  • it would be great if "diff" can have an option to ignore "empty node" which can shift internal IDs even if the graphs are almost same.

RDF converters

• general:
  • RDF --> JSON (for  CoachDB,  mongoDB etc.)
• biological:
  • Bio DB entries -> RDF (in  BioRuby for  TogoWS, for example)
  • OBO <-> RDF (used in  Cell Cycle Ontology because ".obo" file format is easy to read for human)
• Bio2RDF converters:
  • Perl  http://bio2rdf.svn.sourceforge.net/viewvc/bio2rdf/trunk/src/util/perl
  • Python  http://bio2rdf.svn.sourceforge.net/viewvc/bio2rdf/trunk/src/util/python/

Interface for SPARQL endpoint

DBI like interface to Query and obtain Result by SPARQL.

• automatically map to the relevant language object
• we can use SPARQL endpoint as a data source for  Galaxy, for example.
• sometimes, results can be huge

Language availability

language	generic	4store
Ruby	o	o
Python	o	o
Perl	?	x
Java	o	o
R	?	?

• 4store
  • database server  http://4store.org/
    • install doc in Japanese  http://lifesciencedb.g.hatena.ne.jp/nakao_mitsuteru/20100105/1262713627
    • binary download for mac  http://dl.dropbox.com/u/152468/4store-1.0.2.dmg
    • 4storeQuickPrimer
  • client library for Ruby/Python/Java  http://4store.org/trac/wiki/ClientLibraries
• Virtuoso
  • database server  http://virtuoso.openlinksw.com/dataspace/dav/wiki/Main/
    • ->  http://sourceforge.net/projects/virtuoso/files/

Try "short read archive (metadata) SPARQL endpoint" which will be developed by the NextGenSeq? open space discussion group.

Galaxy integration

• Plugin development
• Documentation

Day2

RDF manipulation tools and library to produce RDF from biological objects are important for data providers, but most users are consumers of the data, so we will focus on how to access public biological SPARQL endpoint easily in each Open Bio* library.

Suggested endpoint to play with:

•  Kegg on Bio2Rdf
•  Gene on Bio2Rdf
•  BioGateway endpoint
• More info at:  Semantic Systems Biology

Implementations:

• Python SPARQL client for BioGateway?:
  • Improved:  http://chapmanb.posterous.com/biohackathon-2010-day-4-improved-python-sparq
  • Initial:  http://chapmanb.posterous.com/biohackathon-2010-day-2-python-sparql-query-b
• Initial python query client for InterMine?:  http://chapmanb.posterous.com/biohackathon-2010-day-3-fish-interoperating-a
• Ruby SPARQL client:
  •  ActiveRDF this is the original package, but it seems to be bugged with our endpoints. Two possible solutions:
    • grab  this fork from github
    • apply this patch to the file /var/lib/gems/1.8/gems/activerdf_sparql-1.3.6/lib/activerdf_sparql/sparql.rb (path depends on you configuration)
  • we did tests using Bio2RDF endpoints, we plan to support BioGateway? as well.
  • we can explore the graph dinamically.
  • AGILE NOW!!!!!!!!!!!!! you can see an example of agile sparql on  Raoul's blog.

Todo:

• Survey existing SPARQL library in each language

Todo:

• Can we have common interface for major biological SPARQL endpoints?
• Can we have nice SPARQL query builder?
• Convert retrieved results into language's object?
• What about getting not only a tabular result but also a result displaying a graph (nodes + edges)? (See  BioGateway browser)
• Map directly on Bio* interal objects

Note:

• Half of the BioRuby group will also tackle with to develop the DB (e.g. KEGG) -> RDF generator (will be used in Bio2RDF and TogoWS)