``There are well over 500 public domain data sources of interest to genomics/proteomics researchers. Many of these `data sources' do more than just provide data, they also provide access to a wide range of services. A good example of this are sequence homology search engines. Given the differences in interfaces, syntax and semantics between sites, there is no practical path for a given researcher or research team to use more than a few. Data warehouses, federated systems, and the like help, but only a little. The number of new sources coming online every year, and the number of changes to existing sources, is simply overwhelming. This is one of the major problems driving bioinformatics today.

``We picture a genomics world in which scientists, search engines, and soft-bots can browse and execute (limited) queries against a wide range of sites, with no significant per-site overhead. Rather than attempting to integrate these sources (thus allowing complex queries against few sites), we advocate providing just enough connective tissue to allow semi-intelligent agents or search engines to execute simplified queries against hundreds of sites. The connective tissue can take the form of a collection of loose, service-oriented `schemata' that provide such systems with the information needed to work their way through the interface at each site, to get to the underlying services. A schema might include structured metadata with domain-specific information, a thesaurus, service descriptions, and typical web interfaces. Relevant technology has been demonstrated in the electronic commerce arena.''

URL

http://www-casc.llnl.gov/xewa/

Reference by Snowdeal.org.