Department of Chemistry, University of Florence , Sesto Fiorentino, 50019, Italy.
J Chem Inf Model. 2016 Jun 27;56(6):1117-21. doi: 10.1021/acs.jcim.6b00151. Epub 2016 Jun 10.
We present a new release (6.0β) of the ORAC program [Marsili et al. J. Comput. Chem. 2010, 31, 1106-1116] with a hybrid OpenMP/MPI (open multiprocessing message passing interface) multilevel parallelism tailored for generalized ensemble (GE) and fast switching double annihilation (FS-DAM) nonequilibrium technology aimed at evaluating the binding free energy in drug-receptor system on high performance computing platforms. The production of the GE or FS-DAM trajectories is handled using a weak scaling parallel approach on the MPI level only, while a strong scaling force decomposition scheme is implemented for intranode computations with shared memory access at the OpenMP level. The efficiency, simplicity, and inherent parallel nature of the ORAC implementation of the FS-DAM algorithm, project the code as a possible effective tool for a second generation high throughput virtual screening in drug discovery and design. The code, along with documentation, testing, and ancillary tools, is distributed under the provisions of the General Public License and can be freely downloaded at www.chim.unifi.it/orac .
我们呈现了一个 ORAC 程序的新版本（6.0β）[Marsili 等人，《计算化学杂志》，2010 年，第 31 卷，第 1106-1116 页]，它具有混合的 OpenMP/MPI（开放多处理消息传递接口）多级并行性，专门针对广义系综（GE）和快速切换双消除（FS-DAM）非平衡技术，旨在评估药物受体系统中的结合自由能在高性能计算平台上。GE 或 FS-DAM 轨迹的生成仅使用 MPI 级别上的弱缩放并行方法来处理，而对于具有共享内存访问的节点内计算，则在 OpenMP 级别上实现强缩放力分解方案。FS-DAM 算法的 ORAC 实现的效率、简单性和固有并行性将该代码作为药物发现和设计中第二代高通量虚拟筛选的有效工具之一。该代码与文档、测试和辅助工具一起根据通用公共许可证的规定分发，可在 www.chim.unifi.it/orac 上免费下载。
