Computer Cluster for Computational Biology and Biophysics

 描述：这是一项购买计算机集群的建议，以支持坦普尔大学由NIH资助的六个研究小组(总共有50多名学生、博士后和高级人员)开展跨越生物物理、生物化学、结构生物学、种群遗传学和基因组学的不同项目。具体地说，拟购买一个包含40个计算节点的Linux集群，每个节点有四个16核处理器，总共有2560个处理器核。该集群将提供针对本提案中所述的NIH支持的项目的计算需求进行优化的容量。所需设备将用于的项目包括：蛋白质-配体结合和变构：有效电位、先进采样方法和应用(R.M.Levy)；生物活性分子的渗透和通过分子动力学模拟调节膜通道(M.L.Klein、V.Carnevale和G.Fiorin)；Rosetta程序中新统计计分条件的优化和抗体特异性计算设计的应用(R.Dunbrack)；通过实验和计算方法探索蛋白质折叠的早期阶段(H.Roder和V.Voelz)；基于螺环配基的金属酶模拟物的设计和建模(C.Schafmeister)，以及人类种群基因组统计方法的实施(E.J.嘿)。这些计算密集型项目目前得到7笔R01赠款、1笔P01赠款和1笔P50赠款的支持。所要求的硬件包括具有高吞吐量、低延迟的InfiniBand互连的计算节点和高性能存储设备的平衡组合，以适应需要共享内存节点内并行和松散耦合的消息传递、并行或分离的独立进程在独立数据集上运行的项目。所提出的计算机集群采集的一个关键方面是利用先进的采样算法、膜和离子通道的分子动力学模拟以及用于种群遗传学和基因组学建模的马尔可夫链蒙特卡罗模拟，获得现代蛋白质原子模拟所需的更大的MPI并行能力。本提案中描述的与公共卫生相关的项目包括改进表征蛋白质结构(和组装体)的方法，以此作为治疗艾滋病毒/艾滋病和其他疾病的基于结构的药物设计的基础，以及了解麻醉剂与电压门控离子通道结合的性质。

 DESCRIPTION: This is a proposal to purchase a Computer Cluster to support six NIH-funded research groups at Temple University (in total, more than 50 students, post-docs, and senior personnel) carrying out diverse projects spanning biophysics, biochemistry, structural biology, population genetics and genomics. Specifically, it is proposed to acquire a Linux cluster containing 40 compute nodes with four 16-core processors each, for 2560 processor cores in total. The cluster will provide the capacity that is optimized for the computing needs of the NIH supported projects described in this proposal. The projects for which the requested equipment will be used include: Protein-Ligand Binding and Allostery: Effective Potentials, Advanced Sampling Methods, and Applications (R.M. Levy); Permeation of Biologically Active Molecules and Regulation of Membrane Channels by Molecular Dynamics Simulations (M.L. Klein, V. Carnevale, and G. Fiorin); Optimization of New Statistical Scoring Terms in the Rosetta Program and Application to Antibody- Specific Computational Design (R. Dunbrack); Early Stages of Protein Folding Explored by Experimental and Computational Approaches (H. Roder and V. Voelz); Designing and Modeling Spiroligomer Based Metalloenzyme Mimics (C. Schafmeister), and Implementation of Statistical Methods for Human Population Genomics (E.J. Hey). These computationally intensive projects are currently supported by seven R01 grants, one P01 grant and one P50 grant. The requested hardware consists of a balanced combination of compute nodes with high- throughput low-latency InfiniBand interconnect and a high-performance storage appliance to accommodate projects which require shared memory in-node parallelization and loosely coupled message passing parallel or uncoupled independent processes operating on independent data sets. A key aspect of the proposed computer cluster acquisition is to obtain greater MPI parallel capabilities needed for modern atomic simulations of proteins using advanced sampling algorithms, molecular dynamics simulations of membranes and ion channels, and Markov Chain Monte-Carlo simulations employed in population genetics and genomics modeling. The public health-related relevance of the projects described in this proposal include improving methods for the characterization of protein structures (and assemblies) as a basis for structure based drug design for treatments of HIV/AIDS and other diseases, as well as understanding the nature of the binding of anesthetics to voltage-gated ion channels.