Intelligent Supercomputing

Our intelligent supercomputing platform combines cutting-edge supercomputing technologies, such as GPU-accelerated physics-based simulations, ultra large-scale virtual screening algorithms, and specialized machine learning algorithms to enable rigorous and large-scale pre-screening of protein targets, and chemical compound proposals.

Because our platform enables rigorous dynamics simulations of proteins in fully solvated environments, it excels in the precise characterization and prediction of protein flexibility involved for example in induced-fit ligand binding, and complex solvent-protein interactions that dominate drug binding energetics. In turn, this enables the targeting of previously known but difficult targets, as well as novel or relatively less characterized (e.g., allosteric) flexible binding pockets with higher precision and confidence.

Our physically realistic simulations allow for the virtual estimation of ligand-protein binding affinities through free energy perturbation theory at accuracy levels approaching those of experiment but with a fraction of the cost of actual synthesis and assaying, results in accelerating identification of the clinical candidates.

In addition, our platform includes computational compound screening capabilities at unprecedented scales and efficiencies (e.g., 10 billion compounds per day), enabling evaluation of chemical space that is roughly 10000 times greater than typical experimental high throughput screening campaigns.