There is considerable interest in the Semantic Web as a solution to the growing information technology needs of biological data and drug discovery. However, the progress towards this vision has been very limited. One reason for this slow progress is the difficulty in understanding and applying the concepts of Semantic Web technology to a real-world problem using a rule-based and also semantic dependent URI for machine reasoning. Semantic dependent properties are though definable using ontology, examples of successful rule-based ontology are hard to come by. To circumvent this problem and to facilitate a gradual transition from traditional Web pages to Semantic Web pages for chemical structures with or without a universally accepted ontology, we propose the use of two types of URI; structure invariants that are defined without the use of ontology and structure semi-invariants that are defined using ontology implemented at least locally. Structure invariants are defined by the most fundamental properties (chemical connectivity, bond type, atom type etc.) of the structure. The invariants are used to exchange information on the Web between databases that may or may not use ontology (or if they use one, they are not fully compatible among databases) and they support the exchange of information between the databases in a context independent (non-Semantic) fashion. Structure semi-invariants URI (SURI) are defined from context dependent (semantic) properties of the structure and they support exchange of information using ontology. Binding mode, type of the ring structure, chemical core used in the synthesis of a drug are some of the properties that may be used to define SURI in HIV Structural database (HIVSDB). SURI may be used to exchange information over the Web between databases that use the same ontology. However, SURI may be used, as we illustrate in this paper, to first locally process the user queries using ontology and then to exchange information with other databases using URI. The joint use of structure invariants and semi-invariant URI, as defined above, makes the OWL representation universal (rule-based) and also context dependent (semantic, rule-based for the given context). For modeling using Semantic Web tools, the elements of the OWL model are organized into a chemical taxonomy tree1 (http://xpdb.nist.gov/hiv2_d/advanced_query_files/slide0002.htm ). We model knowledge implied within an established, widely used information resource implemented as a relational database (RDB), the HIVSDB, using OWL (Web Ontology Language2), a standard language of the Semantic Web. We describe this effort, discusses our observations on the model, utility and implementation of Semantic Web technology. We illustrate how the proposed Semantic Web concept may be used to exchange information between four major independently maintained databases; HIVSDB, the PDB, AIDSDB and PubChem. The OWL models for AIDS inhibitors are available at the W3C Web site (http://esw.w3.org/topic/HCLS/ChemicalTaxonomiesUseCase) maintained by the Healthcare and Life Sciences Group of the World Wide Web Consortium. The style of the presentation will be chosen so as to cover a broad audience including Structural Biologists, Medicinal Chemists, and Information Technologists.
Citation: Proceedings of the IEEE
Pub Type: Journals