MRI: Acquisition of a High-Performance Cluster to Enable Advanced Bioscience and Engineering Research

该项目获得一个高性能计算(HPC)集群和一个辅助并行文件系统，旨在满足由于生物和生物技术数据的规模和复杂性增加而导致的计算密集型数据分析和模拟的需求。基础工作解决了在分析、管理和关联由该州跨学科分布式中心(BioSNTR：生化时空网络资源)生成的此类信息时产生的挑战。HPC集群支持BioSNTR的数据分析研究活动，该活动横跨基因组学、光学显微镜、卫星成像、计算化学以及电气和机械工程等学科；事实上，它将成为BioSNTR的枢纽，BioSNTR是由NSF资助的另一个项目创建的全州分布式研究中心，该项目已在整个SD产生了10名新的终身教职跟踪员工。该中心获得了能够产生海量数据的新光学显微镜，需要密集的计算重建方法，从而推动了对当地高性能计算资源的迫切需求。之前的核磁共振成像使下一代基因测序设施得以创建，再加上BioSNTR在生物信息学领域的招聘，推动了对一系列基因组应用的扩展HPC能力的需求。BioSNTR的目标是将先进的成像与下一代测序联系起来，将系统级的基因型与表型联系起来。HPC平台是实现大数据集集成的关键。此外，南达科他州图像处理实验室(IPLab)将受益于该设备。工程应用包括卫星和流体动力计算。IPLab开发了新的技术来解释大气异常，这些异常对从陆地特征测量的光谱反射特征产生了不利影响。未来的工作包括对陆地卫星场景进行广泛的计算处理，以进行气候学分析(300万至400万个数据集)。估计表明，此工作将在一整年内消耗当前集群内的所有处理能力。能源电力系统效率的模拟在很大程度上依赖于高并行处理。生物科学和生物技术已经进入了一个新的时代，在这个时代，新的仪器改变了生物数据的复杂程度。这些仪器产生关于生物表型/功能和基因型的大量信息，从而在分析、管理和关联这些信息方面产生了计算挑战。除了更传统的学科，该仪器还将促进新兴领域的研究。例如，分析经济学中高频金融交易的数据密集型研究方法，以及使用数据挖掘技术研究导致协作对话趋势的社会因素的特征的新闻研究。广泛的影响：该工具允许将并行计算机处理纳入生物、化学和统计科学的研究生课程，并促进计算机科学课程的发展。新的集群利用了最近资助的REU网站，专注于部落和社区大学的早期本科生。调查人员能够与BioSNTR合作，协调每年一次的转录学和图像分析培训讲习班。这些课程将向全州的学生、研究人员和教职员工开放。

This project, acquiring a High Performance Computing (HPC) cluster and an ancillary Parallel File System, aims to be able to fulfill the demand for computationally intensive data analysis and simulation resulting from the increased scale and complexity of biological and biotechnical data. The underlying work addresses the challenge created when analyzing, managing, and relating this type of information generated by the interdisciplinary distributed center in the state (BioSNTR: Biochemical Spatio-Temporal Network Resource). The HPC cluster supports the BioSNTR's data analysis research activities, which spans the disciplines of genomics, optical microscopy, satellite imaging, computational chemistry, and electrical, and mechanical engineering; in fact, it will be a hub for BioSNTR, a distributed research center across the state created by another NSF-funded project that has resulted in 10 new tenure-track hires across SD. This center has acquired new optical microscopes that can produce massive amounts of data, requiring intensive computational reconstruction methods driving pressing need for local HPC resources. A previous MRI enabled the creation of a next generation gene sequencing facility that, coupled with BioSNTR hires in bioinformatics, drives the need for expanded HPC capabilities for a range of genomics applications. BioSNTR aims to link advanced imaging with next generation sequencing to connect the system-level genotype to phenotype. The HPC platform is essential for realizing the integration of large data sets. Additionally, the South Dakota Image Processing Laboratory (IPLab) would benefit from this equipment. Engineering applications include satellite and fluid dynamic computation. IPLab develops novel techniques to account for atmospheric anomalies that adversely impact spectral reflectance signatures measured from land features. Future work includes extensive computational processing of Landsat scenes for climatological analysis (3 to 4 million data sets). Estimates indicates that this work will consume all the processing power within the current cluster for an entire year. Simulating efficiencies in energy power systems relies heavily on high parallel processing. Bioscience and biotechnology have entered a new era in which the new instrumentation transforms the scale of complexity of biological data. These instruments produce expansive information about organismal phenotype/function and genotype, thereby creating a computational challenge in analyzing, managing, and relating this information. Beyond the more traditional disciplines, the instrumentation will facilitate research in emerging areas. Examples include data intensive research methods to analyze high frequency financial transactions within economics, and journalism research using data mining techniques to study the characteristics of social factors that contribute to collaborative conversational tendencies.Broader Impacts:The instrumentation allows inclusion of parallel computer processing into the graduate courses within the biological, chemical, and statistical science and enhances the development of curriculum within the Computer Science program. The new cluster leverages a recently-funded REU site focused on early undergraduate students from tribal and community colleges. The investigators are able to coordinate yearly training workshops in transcriptomics and image analysis in collaboration with BioSNTR. These will be open to students, researchers, and faculty throughout the state.