Beowulf LINUX Cluster Computer

描述(申请人提供)：人类基因组计划的一个主要目标是获取遗传信息，以了解并预防和/或治疗人类疾病。最近完成的人类基因组“草案”测序，以及生物技术的最新进展，使用于基因分型的样本得以快速处理，创造了大量的信息来源。然而，分析影响常见疾病易感性的遗传和环境因素的挑战在统计上仍然复杂，计算也很密集。波士顿大学医学院的遗传流行病学家和统计遗传学家在识别与孟德尔简单疾病和复杂特征的风险相关的基因方面取得了很长的成功历史。为了继续取得这一成功，并实施统计遗传学的最新进展，我们要求建立一个由192个英特尔处理器组成的Beowulf级Linux集群。该计算机系统将专门用于遗传分析，包括连锁分析和关联测试。该系统将支持NIH资助的14个项目的研究。尽管主要用户资助的项目的表型和研究人群有很大差异，但将在该系统上执行的计算密集型遗传分析技术有很大的重叠。这一簇将通过扩展家系中的体面关系来计算同一性，通过模拟计算经验p值，并应用复杂的基因与基因和基因与环境的相互作用模型。这一系统得到了强有力的机构支持，政府承诺为全职系统管理员提供资金，并长期支持升级和维护。在这项提案的调查人员中放置这种计算资源将在遗传学领域产生广泛影响，并进一步了解亨廷顿病、阿尔茨海默病、肥胖、中风和止血因素、认知衰退、高血压、心血管特征以及可卡因和阿片类药物依赖等遗传学问题。

DESCRIPTION (provided by applicant): A major goal of the human genome initiative is to access genetic information to understand and to prevent and/or treat human disease. The recently completed "draft" sequencing of the human genome and the recent advances in biotechnology that allow rapid throughput of samples for genotyping have created a vast resource of information. However, the challenge to dissect the genetic and environmental factors influencing susceptibility for common diseases remains statistically complex and computationally intensive. The genetic epidemiologists and statistical geneticists at the Boston University Medical Campus have a long history of success in identifying genes associated with risk for both simple Mendelian diseases as well as complex traits. In order to continue this success and to implement the most recent advances in statistical genetics, we are requesting, a Beowulf class LINUX cluster of 192 Intel processors. This computing system will be dedicated to genetic analysis including linkage analyses and association testing. The system will support the research of 14 NIH funded projects. Although the phenotypes and study populations differ widely among the funded projects of the major users, there is substantial overlap in the computationally intensive genetic analysis techniques that will be performed on this system. This cluster will enable computation of identity by decent relationships in extended pedigrees, computation of empirical p-values via simulation, and application of complex gene by gene and gene by environment interaction models. There is strong institutional support for this system evidenced by the administration's commitment to funding a full time systems administrator and long term support of upgrades and maintenance The placement of this computing resource among the investigators of this proposal will have a broad impact in the field of genetics and further our understanding of the genetics of Huntington's Disease, Alzheimer's Disease, Obesity, Stroke and Hemostatic Factors, Cognitive Decline, Hypertension, Cardiovascular traits, and Cocaine and Opioid Dependency among others.