Chemists have largely learned from experiments, have drawn generalization and conclusion from their observations to plan new experiments. The large amount of reactions performed in combinatorial chemistry offers great potential for enhancing our knowledge on the scope and limitations of reaction types. A set of reactions performed under identical conditions provides information that has to be stored and systematically analyzed.

The unsupervised inductive learning techniques embodied in CORA can provide insights into chemical reactions, a knowledge that can be used both for synthesis design and reaction prediction. WODCA provides a set of tools for planning a combinatorial library. Foremost are methods for defining strategic bonds, a process that can benefit from the result of a CORA study. Substructure searches can then be utilized to extend the list of starting materials to be used for the combinatorial chemistry experiment. WODCA has been designed to be highly interactive in order that the chemist/her can bring in her/his special interests.

EROS, on the other hand, provides a framework for the simulation of chemical reactions. It needs knowledge derived from a series of experiments to define the scope and limitations of a reaction type, a knowledge again obtainable through the use of CORA.

Thus, both WODCA and EROS trace back their inherent knowledge from careful analysis of a series of chemical reactions. Reaction databases play an important role in this endeavor. However, it must be observed that reaction databases still have large deficiencies: quite often not all information on reaction conditions, or the outcome of a reaction (co-products, by-products, yields) are stored. Thus, we attach large expectations to combinatorial chemistry, hoping that the reactions performed are carefully analyzed and this information is stored in electronic form. This would greatly enhance our potential to learn more about chemical reactions.

0 0

Post a comment