Pack: Supporting material for Roos et al., BMC Bioinformatics, SWAT4LS supplement (2009)

This workflow is for demonstration purposes only. Please contact the authors if you wish to try it. We will gladly collaborate with you.

Summary

This workflow extracts proteins and protein relations from Medline. Extracted protein names (symbols of at least 3 characters) are validated against mouse, rat, and human UniProt symbols, so the results are limited to these species. This workflow follows the following basic steps:

1. it retrieves documents relevant for the query string
2. it discovers proteins in those documents, that are considered relevant to the query string and related to the proteins mentioned in the query (colocation in text mining jargon)
3. it stores the results in a semantic repository

To support hypothesis formation, the results are added to a repository containing proto-ontologies with biological classes and procedural classes to log evidence. The models are based on RDF and OWL.

Acknowledgements:

Synonyms and Uniprot services: Martijn Scheumie, BioSemantics Group, University of Rotterdam, The Netherlands (BioRange project)

Known issues

Occasionally the workflow will fail on intermediate results that return no results (e.g. on a time out or a bug in the workflow). This problem will be addressed in Taverna 2 using its more strict list iteration mechanism and the AIDA plugin for Taverna 2. The workflow contains some elements that are not yet functional. This will show as failed when run. This can be ignored.

Please contact us if you have any questions about the workflow, our approach, or if you experience technical difficulties.