Indexing pharmacogenetic knowledge on the World Wide Web.

A key challenge for pharmacogenetics is the creation of databases to store, analyse and disseminate important datasets in order to catalyse research and training. Most successful databases have a limited scope: Genbank contains DNA sequences [1]; the Protein Data Bank contains the three-dimensional coordinates of macromolecules [2]; the Online Mendelian Inheritance in Man contains a record of human genetic disease [3]; and PubMED contains the biomedical literature [4]. This limited scope is a great strength, because the information can be stored, searched and analysed using a few powerful tools, and the users of these databases know exactly what to expect. Databases for pharmacogenetics and pharmacogenomics will have much more diversity. Pharmacogenetic data involve phenotypes that are as diverse as the assays we invent to measure them. Thus, it is unclear what a user should expect from a pharmacogenetics database, and yet a public repository of pharmacogenetic data is critical to establish a core dataset for the field upon which we can build new analyses and new hypotheses [1]. Clearly, successful databases for pharmacogenetics must employ some sort of classification of phenotypes that is general purpose, yet extensible to include undefined characterizations of phenotype.