Abstract
F. Raymond Salemme
DirectedDiversity¨ process control technology for combinatorial
chemistry uses computer methods to design, select and iteratively refine
combinatorial libraries of small molecule drugs from SAR data or 3D
structural or pharmacophore models. A key feature of the process is the
initial computer generation of very large "virtual" libraries of
synthetically accessible compounds which are designed to explore specific
structural features suggested from an input SAR or structural model.
Virtual libraries typically contain 1,00,000 to 1,000,000 compounds where
each compound has a validated chemical synthesis pathway, and is
characterized by an extensive set of 120+ molecular descriptors. Computer
codes are then used to select library sub-sets (100-1000 compounds) for
rounds of automated synthesis and testing using high throughput bioassay
technology. Test data resulting from compound bioassay during each round
are interpreted by a computer selector code that is able to optimize
multiple objectives simultaneously, in order to refine the properties of
molecules selected in further rounds of synthesis and testing. The
technology has been implemented through a hierarchical client-server system
that tracks compounds from virtual conception through to testing and data
base property registration. Applications of DirectedDiversity¨ technology
to the generation of nanomolar potent protease inhibitors will be
described.
