Bio2RDF is an open source project to generate and provide Linked Data for the Life Sciences. Here, we report on a third coordinated release of ~11 billion triples across 30 biomedical databases and datasets, representing a 10 fold increase in the number of triples since Bio2RDF Release 2 (Jan 2013). New clinically relevant datasets have been added. New features in this release include improved data quality, typing of every URI, extended dataset statistics, tighter integration, and a refactored linked data platform. Bio2RDF data is available via REST services, SPARQL endpoints, and downloadable files.
Bio2RDF is an open-source project to transform the vast collections of heterogeneously formatted biomedical data into Linked Data [1], [2]. GitHub-housed PHP scripts convert data (e.g. flat files, tab-delimited files, XML, JSON) into RDF using downloadable files or APIs. Bio2RDF scripts follow a basic convention to specify the syntax of HTTP identifiers for i) source-identified data items, ii) script-generated data items, and iii) vocabulary used to describe the dataset contents [1]. Bio2RDF scripts uses the Life Science Registry (http://tinyurl.com/lsregistry), a comprehensive list of over 2200 biomedical databases, datasets and terminologies, to obtain a canonical dataset name (prefix), which is used in the formulation of a Bio2RDF URI http://bio2rdf.org/{prefix}:{identifier} and identifiers.org URI. Each data item is annotated with provenance, including the URL of the files from which it was generated. Bio2RDF types and relations have been mapped to the Semanticscience Integrated Ontology (SIO)[3], thereby enabling queries to be formulated using a single terminology [4]. Bio2RDF has been used for a wide variety of biomedical research including understanding HIV-based interactions [5] and drug discovery [6].
Here, we report an update to the Bio2RDF network, termed Bio2RDF Release 3, and compare the results to Bio2RDF Release 2. 2
We redeveloped the Bio2RDF linked data platform to provide 3 basic services (describe, search, links) by querying the target SPARQL endpoint using Talend ESB, a graphical Java code generator based on the Eclipse framework. The REST services now return RDF triples or quads based on content negotiation or RESTful URIs of the form http://bio2rdf.org/[prefix]/[service]/[format]/[searchterm]. The describe service returns statements with the searchterm as an identifier in the subject position. The links service returns triples with the searchterm as an identifier in the object position. Finally, the search service returns triples containing matched literals. Datasets and available services descriptions are stored and retrieved by the web application using a new SPARQL endpoint (http://dataset.bio2rdf.org/sparql). 4
Bio2RDF is accessible from http://bio2rdf.org. Bio2RDF scripts, mappings, and web application are available from GitHub (https://github.com/bio2rdf). A list of the datasets, detailed statistics, and downloadable content (RDF files, VoID description, statistics, virtuoso database) are available from http://download.bio2rdf.org/current/release.html . Descriptions of Bio2RDF datasets and file locations are also available from datahub.io . 5