Supercomputers and Drug Development

A New Model for Pharmaceutical Research 102 91 ndum 10 platir 77 hass 108 78 Pt 192 22 16 Supercomputing-based modeling may help validate and accelerate drug research. 953 % of medicines studied in humans during clinical trials fail to be both safe and effective. Mission: Streamline drug discovery Drug companies often devote 10 to 15 years Millions of drug candidates and more than a billion Drug diecovery and high- throughput screening dollars to bring a new drug to market. Structure-based drug design Computational chemistry efforts at the Lawrence Livermore National Laboratory may help streamline the drug discovery process and enable researchers to bring new therapies to clinical trials and the Drug metaboliem and toxicity teste Includes in vitro and animal tests Clinical triale Drug approval marketplace more rapidly and with a higher rate of success. NE 25MG CAP GLE ONE OR TWO BEFORE BEDT Screening drug candidates ... with supercomputers Desktop computer High-performance computer 10 years 1 year 1 month 1 day 1 hour 100 CPUS 15.000 CPUS 700 CPUS 1 CPU 1 minute Prepare target Prepare molecules Run VinaLC docking Rescore Tasks High-performance computing (HPC) systems can drastically reduce the current virtual screening time frame and increase the feasibility of more accurately screening extremely large compound databases. In this example, researchers used the Livermore code VinalC to test about 40,000 molecules against a single drug target. This technique can help to efficiently reduce the list of drug candidates to a manageable number for synthesis and testing. (2 Predicting molecular behavior of drugs Drug molecule Lipid barrier This simulation shows a drug compound crossing the lipid barrier between the bloodstream and the brain. Predicting such behavior could help drug researchers better gauge drug efficacy and catch potentially serious side effects at an early stage of drug development. 3) Validating models with public data Vascular Tumors Immune Blood and Psychiatric Endocrine Renal and lymphatic Liver Gastro- Cardiac disorders disorders system disorders disorders disorders disorders intestinal disorders disorders urinary disorders Adverse drug-reaction categories A Livermore-developed virtual adverse-drug-reaction screening program integrates in-house molecular data with publicly available data sets. Models tested with Livermore's VinaLC calculation (red) and with experimental data (blue) are roughly comparable at predicting adverse-drug-reactions. This graph illustrates model quality averaged across 10 rounds of cross validation for 10 categories of serious or lethal reactions, for a total of about 500 drugs. (4) Predicting how drugs will be Site 1 metabolized When breaking down a drug, the body may produce one or more by-products, called metabolites. Some of these metabolites may be toxic, even when the drug itself is not. Livermore researchers calculated the most favorable binding arrangements for several key proteins during acetaminophen metabolism. The Site 2 free-energy mapping (right) and binding calculations (insets) tell us that proteins prefer to bind at site 1, but that this arrangement can create a by-product that can cause liver damage. Binding at site 2 would produce a nontoxic by-product; however, it's less likely. 5 Modeling a drug's behavior as it flows through human systems Acetaminophen also served to validate a Lung study that examined drug metabolism and adverse Brain reactions on a larger scale. Bone In Livermore's 14-compartment physics-based kinetic model of the Heart human body, Muscle various tissues are connected by the circulatory system, and key processes in the human response to drugs are included. This Kidney Spleen Liver Gut model will help researchers assess Skin toxicity risks early in the drug development Body fat process. Rest of body Lawrence Livermore National Laboratory www.llnl.gov @Livermore_Lab Read more in Science & Technology Review https://str.lnl.gov/july-2014/lightstone Venous blood Aoenoce jopow More accurate - ogeunooe sso CHEMISTRY Arterial blood