The de novo design of novel compounds that are bioactive, have good ADME properties and are synthetically accessible is extremely challenging. We have developed a knowledge-based tool that addresses synthetic accessibility by encoding the transformations that occur in known reactions into reaction vectors. The reaction vectors can then be applied to previously unseen starting materials to generate new products. The reaction vectors are derived automatically from any collection of reactions and therefore allow the generation of large and comprehensive knowledge-bases. The vast size of chemical space, even when design is restricted to relatively small numbers of reactions, is such that a highly efficient search strategy is required. We have integrated the structure generation step into a multiobjective optimization algorithm with the goal of generating novel molecules that are optimized on a number of design objectives. We demonstrate our method on design scenarios in fragment-based drug design.

De novo design of synthetically accessible compounds: Application to fragment-based drug design

243rd National Meeting (2012)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.