SciNetS Search : Inference Search over an Integrated Life-sciences Database Based on the Semantic Web

摘要

Integration of various omics databases is a fundamental approach for comprehensive understanding of life phenomena. Scientists’ Networking System (SciNetS) is a Semantic Web-based data repository that realises database integration by collecting metadata of databases described as a set of triples consisting of two bio-items such as gene and ontology terms and a semantically defined relationship between these bio-items. These triples are curated from various life-sciences databases including gene-publication relationships manually curated from MEDLINE and gene-phenotype relationships extracted from published databases by computer programs. SciNetS internally forms a metadata network by maximally concatenating triples. SciNetS Search realizes a cross-search over the integrated databases by traversing the metadata network managed by SciNetS. SciNetS Search allows a user to discover bio-items included in the ontological category specified by the user. The results include bio-items not only having a user’s keyword but also semantically inferred from these bio-items using metadata relationships, and are ranked using statistical scores computed on the basis of the metadata network. SciNetS Search is available via a user’s web browser and provides search service against over 30 million bio-items of 100 thousand categories and 100 million relationships curated from over 100 public omics databases of various species including human, mouse, rice and _Arabidopsis_.