Combining terminologies and ontologies to integrate biomedical information

The post genomics era is characterized by huge amounts of biomedical information, distributed in multiple databanks (e.g. SWISS-PROT, OMIM, LocusLink, GenBank, as well as many others). Despite recent efforts to provide standard ontologies such as Gene Ontology, semantic heterogeneity is a major obstacle to information integration. Each databank has its own identifiers for genes and gene products; the names of biological entities are associated with synonymy and ambiguity; there are numerous biomedical terminologies in use (e.g. MeSH for indexing biomedical literature). Therefore, besides the needs for ontology, there are needs for various mappings (e.g. between terminologies) and cross-references between databanks. This paper presents an ongoing project, BioMeKE that aims at developing an information integration system providing a unified access to biomedical resources. Semantic integration in BioMeKE is based on the combination of existing terminological and ontological resources, including the Unified Medical Language System (UMLS), which integrates sixty families of biomedical vocabularies in a repository of around 900,000 concepts organized according to a set of 135 Semantic Types. Other resources provide synonyms and cross-references, such as the Genew database.