Title: Anton: A Specialized Machine for
Millisecond-Scale Molecular Dynamics Simulations of Proteins
The ability to perform long, accurate, atomic-level molecular dynamics (MD) simulations
of proteins surrounded by explicitly represented water molecules could in principle
provide scientifically (and perhaps pharmaceutically) relevant insights into the structural,
dynamic, and functional characteristics of proteins. A wide range of biologically
significant phenomena, however, occur over timescales on the order of a millisecond—at
least two orders of magnitude beyond the reach of the longest previous MD simulations.
We have recently constructed a working prototype of the first of 16 “segments” of a
specialized, massively parallel machine, called Anton, designed specifically for the
execution of MD simulations. Each Anton segment was designed to simulate a single
MD trajectory as much as a millisecond or so in length using novel algorithms developed
within our lab. The prototype machine very quickly generated the world’s longest (by a
significant factor) MD trajectory, and is now being used within our lab for preliminary
tests involving the simulation of a number of proteins.
This talk will provide an overview of our work on parallel algorithms and machine
architectures for high-speed MD simulation, and will describe research conducted within
our lab in which simulations of moderate length have helped elucidate the dynamics and
functional mechanisms of several biologically important proteins. Some of these
computational studies have yielded testable predictions that have subsequently been
validated through laboratory experiments.