AbstractsComputer Science

Proteins are the "machines of life" and understanding their motions is crucial to understanding the nature of various diseases, for instance Alzheimer's and Huntington's. Molecular Dynamics simulations are useful since they provide a atom-level resolution of a protein's motion. Of particular interest are measuring the rates at which different molecular conformations interchange. A significant bottleneck in the use of molecular dynamics is the difference in timescales. As a result, simulations require billions of steps to access slow motions such as protein folding. This work describes the development of a software pipeline to address this issue. The result is a set of software packages aimed at facilitating specific points is the study of pro- teins: exploratory simulations, conformational sampling, and rate calculations. Due to the large number of simulations that need to be run, each step supports the use of distributed systems and supports long-running applications and fault-tolerance. problem addressed is that current approaches suffer from systematic bias and my contribution is an implementation of a method that does not suffer this bias, as well as a distributed computing pipeline.