Shortcuts For Building Better Molecules
Sandia National Laboratories is doing computer searches for new pharmaceuticals using software that was originally developed to help arrays of telescopic video cameras detect incoming ballistic missiles. New York-based Sterling Winthrop Inc., the drug subsidiary of Eastman Kodak Co., is Sandia's partner. By sifting out promising compounds electronically, Sterling hopes to quickly home in on classes of molecules that merit lab experiments. That could slash the typical $50 million cost of synthesizing and testing compounds for a new drug, says John Wendoloski, director for biophysics and computational chemistry at Sterling.
Sandia's contribution is a trick that quickly narrows complicated searches to a few promising candidates. When hunting for the best molecular shape to fit a particular protein, there are billions of possibilities. Trying them all could take decades, even with a supercomputer. So researchers at Sandia's Livermore (Calif.) lab are harnessing a so-called genetic algorithm that has been tuned to root out optimal solutions to complex problems. The work is still in its early stages. Under terms of their Cooperative Research & Development Agreement, Sterling will have exclusive use of the Sandia software for two years, after which it could be licensed to the public.