资 源 简 介
An important component of rational structure-based molecular design is the generation of large ensembles of conformers for small molecules of interest. Such ensembles are used as input to score and evaluate binding interactions in docking engines, as well as to maximize host-binding affinity. There are two general approaches to conformer generation: systematic or stochastic. In this work we leverage the increasing number of processors and speed of high-performance computer clusters to increase the ability and accuracy of systematic conformer generation and approach the limit of an exhaustive search of the conformational space. The algorithms developed here approach linear scaling and address issues that plague high processor count jobs, such as limited memory, intercommunication overhead, and load imbalances. The ideas presented are used to improve conformational searches of small molecules that are currently outside the domain of available software. Further, we explore the
