BioMap: gene family based integration of heterogeneous biological databases using AutoMed metadata

This work presents an extensible architecture that can be used to support the integration of biological data sets. Biological research frequently requires this kind of synthesis. However, the data models on which biological data sets have been constructed are heterogeneous and difficult to use together. Our architecture uses the AutoMed data integration toolkit to store the schemas of data sources, together with the transformation from these schemas into a global integrated schema. The transformation encompasses two parts; the incremental construction of a global schema which unifies the various data source schemas, and the identification of semantically identical labels for entities. Entities in the unified resource are integrated using PFS-cape. This categorises the entities into clusters based on sequence similarity, allowing the use of family information in the annotation of expression data and experimental target selection.