Searching Composition Space in Combinatorial Chemistry by Monte Carlo Methods
BACKGROUND: Combinatorial chemistry involves searching a large compositional space for compounds with a high figure of merit in order to find a molecule with a given property. However, current methods of searching only apply to relatively small libraries of compounds, with limited numbers of compositions and figures of merit that change smoothly with the composition. For example, exhaustive search methods fail when the potential composition space is larger than can be constructed or searched in a single library, and the genetic algorithm approach tends to fixate on local optima instead of finding the best set of molecules.
DESCRIPTION: Researchers at the University of California have developed an improved method of finding optimal molecules based upon Monte Carlo simulation. In this combinatorial strategy, a library of N molecules is made and the desired property of each molecule is measured. A new library is then constructed and the process is repeated. The synthesis and screening would be under robotic control. The approach is suitable for large composition spaces and rough figure-of-merit landscapes (i.e., those cases where the combinatorial approach is most helpful).
APPLICATIONS: This invention has applications in combinatorial chemistry.
ADVANTAGES: The new UC technology provides the following benefits:
* Improved method for finding optimal molecules;
* Guaranteed to properly sample the desired figure of merit, which is relevant to patentability, cost of materials, and ease of synthesis;
* Sampling is useful to produce a set of molecules that can be subjected to a secondary screen for such issues as patentability, cost of materials, and ease of synthesis;
* Allows for the complete sampling of a chemical space.
US 6,640,191 [MORE INFO
Type of Offer:
« More Chemistry Patents