面向云环境的蛋白质折叠模拟计算并行化算法

摘要

提出了面向云环境Yarn(yet another resource negotiator)规范的蛋白质折叠模拟计算并行化算法Yarn PERM.分析了蛋白质折叠的格点模型 PERM 算法的运行流程及其面向 MapReduce的子任务划分方式.Yarn PERM 算法实现采用 Ha-doop2.0的Yarn框架作为工作平台,其资源的分配与调度、应用子任务的申请和子任务的具体执行都由Yarn来透明地完成;描述了Yarn PERM算法的Map程序与Reduce程序及主控程序的功能实现.实验结果表明:在相同的时间内Yarn PERM比PERM串行计算、MapReduce的PERM 计算在能量最低寻优的吞吐量上明显增加,加速比和可扩展性上也有明显的优势.%A Yarn oriented protein folding simulation parallel algorithm named Yarn PERM was proposed. The computation flow of PERM algorithm and the subtask partition for MapReduce approach in Yarn PERM were dis-cussed subsequently. The Yarn framework of hadoop2.0 was used for the protein folding working Platform in Yarn PERM. All the Resource distribution and scheduling, subtask applying and subtask executing were completed transparently in Yarn. The Map program and Reduce program for Yarn PERM were also described in detail. In or-der to demonstrate the efficiency of Yarn PERM,the simulation experiments were done. The results and perform-ance analysis demonstrated that the finding lowest energy function had been completed in a more throughput num-bers than PERM and MapReduce PERM. Good speed and scalability had been also obtained.