Semantic Web Atlas Project

RDF Graph Overview

Gene, p value, differential expression, and experimental factor are connected to experiment accession via a blank node and a generic is_about relation in order to associate them in the context of a specific experiment only. The following OBI and EFO predicates were used to model the data:

is_about <http://purl.obolibrary.org/obo/IAO_0000136>
experiment accession <http://www.ebi.ac.uk/efo/EFO_0004033>
gene <http://www.ebi.ac.uk/efo/EFO_0002606>
discretized differential expression <http://www.ebi.ac.uk/efo/EFO_0004034>
p value <http://purl.obolibrary.org/obo/OBI_0000175>
experimental factor <http://www.ebi.ac.uk/efo/EFO_0000001>
organism <http://purl.obolibrary.org/obo/OBI_0100026>

Triples were generated using a custom XSLT transform applied to XML REST output of Gene Expression Atlas queries for all available experiments. Project files released under LGPLv.3 are available from sourceforge.

Triplestore

The total number of triples is ~170 million. The triplestore is kindly hosted by Jun Zhao, funded by the EPSRC LSI Postdoctoral Fellowship, and powered by Jena TDB and Open-Source Software Milarq.

The data can be accessed via the SPARQL endpoint:

http://www.open-biomed.org.uk/sparql/endpoint/gxa

To access a UI for the SPARQL endpoint, please go to:

http://www.open-biomed.org.uk/sparql/query.html

You can run the ASK query to test the endpoint. However, to ensure the quality of the SPARQL endpoint, it does not support by default aggregate queries (COUNT, etc.). If you do need to test such queries, please email us and we will send you the more open access SPARQL endpoint.

A secondary (failover) endpoint running Sesame:

http://wwwdev.ebi.ac.uk/microarray-srv/openrdf-sesame/repositories/gxa

Example SPARQL queries:

http://code.google.com/p/open-biomed/wiki/GeneExpressionAtlas

Contact

Tomasz Adamusiak and James Malone