GlaxoSmithKline Position Paper for Workshop of Ontologies in Agents Systems Robin McEntire GlaxoSmithKline Pharmaceuticals 709 Swedeland Road, PO Box 1539 King of Prussia, PA 19406-0939 610.270.6527 Robin_A_McEntire@gsk.com ABSTRACT a common syntax, which enables the semantic representations to This paper describes the position of GlaxoSmithKline with be moved from one site to another without undo, or inaccurate, respect to the development of ontologies and intelligent, agent- translation of the definitions of the underlying objects. And well- based systems. defined services allow easy programmatic access to life sciences objects and to services that manipulate those objects (and produce other life sciences objects). Keywords ontologies, agents, life sciences. 2. Commercial Opportunities More importantly, over the long run, this combination of well- 1. Motivation defined services and well-understood, easily exchangeable objects It is clear that the number of data sources available and necessary will create a marketplace in which vendors may compete to to move forward in the study of the life sciences is quite large and provide services and new data sources. As we have seen in other diverse and is growing at a remarkable rate. It is also clear that technology areas the greatest driver for the adoption and use of a there is tremendous growth in the field in computational methods technology is a strong marketplace. GSK, and other pharmas, are and services for manipulating this data, often manipulating interested in services that provide scientific information that can multiple, disparate, distributed and heterogeneous sources of data. be easily geared to the specific objectives of its scientific The proliferation of data formats and schemas for life sciences investigators. So, for example, there is currently work being done objects and the many methods for accessing or computationally to provide more targeted access to scientific journal articles for manipulating those objects and services is a serious impediment to the research scientist by many of the electronic journal vendors. the efficient, effective and scalable use of these data and services. This process could be significantly aided by categories and The problem will become only more severe as the field continues ontologies that are shared across the industry. The workshop to grow. However, the recent developments in web technology, organisers have already suggested that one of our goals should be including web languages, and object-oriented and knowledge- finding applications that provide the greatest internal business representation technology, are at a level of maturity that they can impact or "the most bang for the buck", which is a laudable goal. be leveraged to help solve this problem. However, we should also consider the markets that our work might target and what opportunities there might be for us to As outlined by the workshop organisers a progressive framework leverage our work into new or existing COTS products. By for the life sciences has a number of components, which are; focusing on the greatest external business impact this work will 1. the development of common, shareable ontologies for objects yield the greatest impact on our industry as a whole. within the life sciences domain 2. the development, or, more likely, the adoption of one or a 3. Additional Technologies few common languages for the exchange of these common In addition, GSK is very interested in intelligent integration of ontologies, and multiple, heterogeneous data sources. Well-defined services and 3. the definition of services that provide well-characterised, common object definitions will make the job of mediated access uniform (or, at least, consistent) access to data/information to multiple data sources much easier. It will also make the display and services and visualization of this information more intelligent and geared more to the particular interests of each scientist. There are already Common, shareable ontologies provide the field with an some vendors in the market and others emerging. Our efforts understandable semantics for discussion and for the programmatic could provide substantial assistance to these vendors. The same computation of life sciences objects. Exchange languages provide could be said of text mining software and products. We are seeing a number of vendors emerge in this area. Ontologies and related services could be a significant aid here as well. 4. Relationships and Synergies We also encourage the organisers of the workshop to consider what other organisations might provide in terms of valuable synergies for our work. For example, the OMG has been in existence for a considerable time and has been very successful in defining common services within the life sciences research Encyclopedia of E. coli Genes and Metabolism. Nuc. Acids domain. Another significant strength of the OMG is that it's Res., Vol 27 No 1, pp 55-58. membership includes a large number of IT and Life Sciences Karp, P.D., Chaudhri, V, XOL Specification, organisations, who now have a history of working together on a http://www.ai.sri.com/pkarp/xol/. number of problems. McEntire, R, et al, An Evaluation of Ontology Exchange Languages for Bioinformatics, In Proceedings of ISMB 5. References 2000. Cattell, R.G.G., et al, 2000. The Object Data Standard: Object Management Group, 1997. UML Semantics, Version ODMG 3.0, Morgan Kaufmann Publishers 1.1, ftp://ftp.omg.org/pub/docs/ad/97-08-04.pdf. Fikes, R., Kehler, T. 1985. The Role of Frame-Based Riley, M., 1993. Functions of the gene products of Representation in Reasoning Communications of the Escherichia coli, Microbiological Reviews, 57:862-952. Association for Computing Machinery, 28(9):904-920. Rumbaugh, J., Jacobson, I., Booch, G., 1998. The Unified Gruber, T.R. 1993. A translation approach to portable Modeling Language Reference Manual, Addison-Wesley. ontology specifications. Knowledge Acquisition, 5:199-220. Karp, P., Riley, M., Paley, S., Pellegrini-Toole, A., Krummenacker, M. 1999. EcoCyc: Electronic