Supercomputing for Pharmaceutical Research



The emergence of multidrug-resistant bacteria in recent years highlights a growing problem in the drug industry—the need for novel antibiotics and the challenge of developing them. Livermore computational biologist Felice Lightstone and her team have researched a possible solution—high performance computing modeling that may accelerate the development of medical countermeasures such as antibiotics. In a collaboration with Trius Therapeutics, Inc., the team demonstrated how computer-based screening can minimize costly laboratory experiments and help shorten the chemistry phase of drug development from three or four years to six months, a vast improvement over conventional testing methods.

With funding assistance from the Lab-Directed Research and Development Program, Lightstone and colleagues also turned their attention to another area—enumerating adverse drug reactions (ADRs) of a particular drug and zeroing in on which drug candidates are likely to be effective during clinical trials. A computational technique known as virtual screening is traditionally used to identify which candidate drugs contained in publicly available databases are most likely to bind to a targeted receptor, which can effectively reduce the list of drug candidates to a more manageable number. In order to expand the capability of virtual screening so that it could be scaled to high performance systems, Livermore researchers optimized a popular docking program for parallel computing, VinaLC. The result was a drastically reduced screening time with the potential to evaluate even larger pools of candidate drugs. While science is nowhere near uploading a person’s genome into a computer program to obtain a tailored therapy, the research has major implications for speeding up the drug development process and predicting the outcome of therapeutic treatments.

See the full S&TR article.