EAGER: Multi-Domain, Workflow Driven Computation System for Microbial Ecology Research and Analysis

这个EAGER奖给了加州-圣地亚哥大学，探索了先进微生物生态学研究和分析的社区网络基础设施的创新，（CAMERA，http：//camera.calit2.net/）。 CAMERA是一种广泛使用的基础设施，它提供了一个单一的软件系统，用于存储，定位，分析，可视化和共享有关微生物生物学的数据。 CAMERA的一个被大量使用的核心组件是用于数据分析的工作流驱动计算系统。 该奖项预计下一代序列（NGS）数据所需的分析。 BLASTx工作流程（使用CAMERA的Refseq蛋白质数据库）将需要近250万核心小时来完成对典型NGS数据集的分析，这使得依赖单个计算资源进行整个工作流程变得不切实际且成本高昂。 PI将原型化新的基于并行计算的动态资源调度方法，以允许工作流任务在异构计算域之间的分布，例如本地系统、各种商业云和多个国家资源。本研究探索了多域调度系统的开发，其中工作流的各个组件任务不绑定到单个计算资源或域，而是可以跨多个资源以及跨机构域来调度。 它使大型和小型实验室中快速增长的科学界部分受益，这些实验室的仪器现在以适度的成本产生大量的基因组数据，但他们无法获得处理或分析这些数据所需的资源。 即将建立的网络基础设施将有可能被面临类似计算挑战的其他科学界使用。

This EAGER award to the University of California - San Diego, explores innovations to The Community Cyberinfrastructure for Advanced Microbial Ecology Research and Analysis, (CAMERA, http://camera.calit2.net/). CAMERA is a widely used infrastructure that provides a single software system for depositing, locating, analyzing, visualization and sharing data about microbial biology. A heavily utilized and core component of CAMERA is the workflow driven computational system for data analysis. The award anticipates analysis needed for next generation sequence (NGS) data. BLASTx Work flows using CAMERA's Refseq Protein database) will require nearly 2.5 Million core hours to complete analysis of a typical NGS data set, making it impractical and costly to rely on a single computational resource for an entire workflow. The PI will prototype novel heuristic-based dynamic resource scheduling approaches to allowing distribution of workflow tasks among heterogeneous computing domains such as local systems, a variety of commercial clouds, and multiple national resources.This research explores development of a multi-domain scheduling system where the individual component tasks of a workflow are not bound to a single computational resource or domain, but rather can be scheduled across multiple resources as well as across institutional domains. It benefits a rapidly growing segment of the scientific community in both large and modest laboratories whose instruments now generate massive amounts of genomic data at modest expense but who have insufficient access to resources needed for processing or analyzing these data. The cyberinfrastructure to be established will potentially be usable by other science communities facing similar computational challenges.