Visualization

We have focused on the development of interfaces to explore and visualize Semantic Web knowledge bases. We have focused on RDFScape, a plugin that allows to integrate Semantic Web knowledge basis in Cytoscape. The objectives for this hackathons were:

• to make the interface more user friendly
• to improve robustness
• to collect application cases and demo.

During the hackathon, the interface has been re-designed and large part of the code have been rewritten to automatize many tasks and reduced the complexity exposed to the user. In this re-implementation, the plugins allows Cytoscape to be used either as:

• an interface to query and visualize remote sparql endpoint
• a tool with its integrated trisplestore (not over sparql!)
• a tools for interactive analysis and reasoning on ontologies and RDF models.

These three operation modes have been partioned in order to simplify the user experience in specific tasks. The general robustness and the quality of the user experience has greatly improved.

The re-design of the plugin is not finished yet, but almost 50% of it has been completed. Most notably most of the parts that required joint cooperation with the Cytoscape team.

We didn't have the time to proceed with a collection of use cases, though we hope these will come from participants when the plugin is fully operational.

Francois also illustrated the use of RelFinder?, which is promising, under specific cirumstances (performant endpoint and "local enough" selections)

MOLECULAR INTERACTIONS

We have been working in trying to make PSICQUIC output RDF/XML. PSICQUIC is a standard that defines a common way to access Molecular Interaction resources (IntAct?, MINT, BioGrid?, iRefIndex, MatrixDB, Reactome x2, MPIDB, ChEMBL), reaching more than 1.7 milion interactions.

 http://www.ebi.ac.uk/Tools/webservices/psicquic/registry/registry?action=STATUS

The approach taken has been to try to make PSICQUIC services return BioPAX Level 3 data. As the underlying data is PSI-MI XML, a converter has been created using JENA.

At the end of biohackathon, PSICQUIC is able to return RDF data in a variety of formats (RDF/XML, N3, n-triples, etc). This may be a feature of the next release of the reference implementation.

At the moment PSICQUIC is returning part of all the information available, but it is a matter of continue implementing the converter to return the rest of data. Currently, the data returned contains interactions, its participants and cross-references for the participants. The ID of the interactions use IntAct? URIs from uniprot (e.g.  http://purl.uniprot.org/intact/EBI-12345), so in the future this will be changed in order to return the corresponding URI for all the different services depending on the original data source.

A test to import the output data into Cytoscape 2.7.0 beta 3 has been successfully done.

ABOUT OTHER PSI FORMATS

PSI-MI XML is one of the PSI (Proteomics Standards Initiative) formats. There are other standard formats to manage mass spectrometry data (mzML), data analysis (analysisML) or sample preparations (spML). These may be converted to RDF equivalent formats in the future. Next month is taking place the PSI Meeting in Seoul, where this may be discussed.