IMPROVING CONFORMATIONAL SAMPLING IN SIMULATIONS OF BIOMOLECULES

点击翻译按钮获取中文摘要

Our research focuses on methods to increase the efficiency of the sampling of conformational space during molecular dynamics (MD) simulations. Such sampling is critical for many applications of MD, specifically including (but not limited to) the calculation of free energy differences due to chemical or conformational changes in a molecule as well as prediction of the three-dimensional structure of biological molecules. We have performed simulations to calculate the anomeric free energy differences for several carbohydrates using the standard methods available in the AMBER suite of programs. Currently, we are implementing the Locally Enhanced Sampling (LES) method in AMBER, and have repeated the calculations using LES, determining the relative benefits than can be obtained as well as the computational overhead. Additionally, we are investigating possible methods to improve the ability of MD simulations to predict the conformation of biological molecules in solution. Most such predictions are severely restricted by the limited conformational sampling that can occur during affordable simulations. we have therefore also implemented the LES method into the MD portion of AMBER, and we are currently running extensive tests to judge the performance. we are using LES both alone and in combination with Simulated Annealing (SA), comparing the results from simulations of protein loops to those obtained by conventional MD simulations. We use the Computer Graphics Laboratory for visualization of the results of our simulations.