Supercomputing in biology: towards understanding living systems in atomic detail

Abstract

Biotechnology has achieved exquisite control over many biological systems. However, to design drugs computationally, we need to understand these systems at the atomic level of detail well enough to predict their behavior. While we have not achieved this goal, advances in high-performance computing are helping considerably. The advent of petaflop-scale computing has opened the door to realistic simulations of biomolecular systems, making it possible to simulate major drug targets, such as the ribosome, in atomic detail. In addition to being the target of approximately half of antibiotics used in US hospitals, the ribosome is also interesting from an evolutionary point of view. Because many regions of the ribosome are identical in every organism studied to date, the ribosome is arguably the oldest molecular fossil, and represents a bridge between the RNA-world and the modern DNA-RNA-protein world. Analogous to the 'CPU of the living cell', the ribosome performs a look-up table operation, converting the 4-letter alphabet of genetic instructions into the 20-letter alphabet of proteins, the cell's workhorse molecules that perform biochemistry and constitute structures of the cell. We have used high performance computing to delve into the inner workings of this molecular factory, applying Newton's equations of motion to every atom in the ribosome and its surrounding environment, iterating for billions of time steps. In addition to revealing mechanistic insight unattainable by experimental studies, these simulations present an excellent system to push the limits of supercomputing - even exa-scale simulations are insufficient to simulate the entire process of protein synthesis.