The search for new drugs to combat disease is an endeavor of growing importance, especially as organisms once thought conquered begin to evolve immunity to existing compounds. For many years, the chief strategy involved finding compounds by screening large numbers of naturally occurring substances for activity. The active compounds were then identified, and synthesized in hundreds or thousands of variations on their basic structure until something acceptable was found. This is often described as the "shotgun" approach.
Attempts to predict biological activity from chemical structure have led to a veritable alchemy of correlations, which continue today under the name of "quantitative structure activity relations", or QSAR. With the advent of reliable methods of determining spatial molecular structure and inexpensive high-speed computers, the field of 3-D structure-based drug design was born. To date, however, the lack of a simple and reliable method of predicting the binding affinity of a "ligand" to a "receptor" protein of known structure (molecular complementarity) has limited its success. More recently, methods based on "combinatorial chemistry" and "test-tube evolution" have been developed, that promise to make it possible to screen large numbers of compounds simultaneously and/or systematically search for those with improved activity. In each of these diverse approaches, significant computational and mathematical problems exist which limit their applicability and effectiveness.
The purpose of this meeting is to acquaint pharmaceutical and computational chemists with the latest developments in such fields as discrete and computational geometry, statistics, data analysis and systems theory, which may prove applicable to the above approaches, while at the same time helping mathematicians and computer scientists to identify the relevant problems.