Presented by T.S. Carpenter for Joint Design of Advanced Computing Solutions for Cancer.<br/><br/>Molecular dynamics (MD) simulations can reveal the atomistic scale mechanisms of biological systems in great detail. However, MD calculations are computationally intensive, and many important biological phenomena are inaccessible because of large time and length scales. To overcome some of these limitations, Pilot 2 of the Joint Design of Advanced Computing Solutions for Cancer in collaboration with International Business Machines Corporation (IBM) are developing a new type of massively parallel, multiscale simulation framework in which machine learning couples a novel continuum macro model to an ensemble of micro MD simulations. The result is simulations at macro length and time scales with microscopic precision and self-adaptation. I will report on the deployment of this scalable framework on Sierra (currently ranked second on the TOP500 supercomputing list) and our results of simulations of RAS proteins at physiological concentration on a (1x1 micrometer) cell membrane.<br/><br/>

Bridging the scales: Machine learning directed macro-to-micro scale simulation to model KRAS initiation of cancer

2019 Fall National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.