Searching BioModels Database

Abstract

*Aim*The increasing number of computational models of biological systems demands a better support for search facilities. One possible improvement is to enable ranked retrieval of models based on existing model annotation. Applying existing Information Retrieval (IR) techniques to computational models, we have implemented a new search strategy in BioModels Database. In this search strategy the index of search terms contains not only the model descriptions in the database, but is expanded by including ontological and textual information about a model and its constituents retrieved from external sources. To add more background knowledge, we additionally index the abstracts of the reference publications. The new search system was integrated into BioModels Database and is based on a Lucene implementation. The model index includes 454 models and 154854 terms, the semantic index 2353 constituents and 437244 terms. Different sources are incorporated in the retrieval and ranking process, the most important of them via MIRIAM annotations to models and model constituents.The search results are sorted with regard to different aspects, i.e. model authors, dates, constituents, or SBML elements. Several state-of-the-art information retrieval features are supported, such as fuzzy, proximity or range search. As a result the user is enabled to specify the criteria relevant for a search, and is provided with a ranked result set of relevant models. The search was implemented in the demo version of BioModels Database and is available at http://www.ebi.ac.uk/biomodels-demo/.A detailed description of the technical realisation is given in:[Henkel et al., 2010]: Henkel, R., Endler, L., Peters, A., Novere, N., and Waltemath, D. (2010). Ranked retrieval of computational biology models. BMC Bioinformatics, 11(1):423+..