Mediated taxonomy system for bioinformatics data integration

In biomedical/biological research fields, scientists usually face the problem of information explosion. Once a query is issued, a collection of data instance IDs such as protein accession numbers are returned. Each protein must be checked for the features of interest. This paper addresses the characterization and differentiating of biological data in the context of mediated domain ontology based on the multilevel abstraction framework. Our mediator system features hybrid ontologies (internal core ontology concept and external classification/annotation concepts) for the interpretation of protein and gene instance data in the context of interaction, pathway and process, which achieve the goal of data integration at the instance level.