GPU-accelerated Peptide Search

描述(由申请人提供)：建议使用图形处理单元(GPU)设备来提高肽搜索的速度多达一个数量级。这项提议是为了在高度并行的GPU设备上使用日益流行的通用计算方法来加速多肽搜索。使用GPU以低成本获得桌面超级计算机的性能是一个活跃而富有成效的研究领域。因此，研究人员将能够进行今天必须在计算机集群上分析的实验，而不是在他们的台式计算机上分析它们。 与公共健康相关：建议对流行的多肽搜索引擎X！Tandem进行增强，使其能够在图形处理单元(GPU)卡上以极快的性能并行运行。这将允许更快、更便宜的操作，使蛋白质组学研究人员能够使用他们现有的台式计算机分析大型蛋白质组实验或搜索翻译后修饰(PTM)。蛋白质组学在生物学理解方面取得了许多重要进展。然而，许多有价值的数据集没有搜索PTM，仅仅是因为进行搜索所需的计算机能力不可用。有了这个项目，我们计划通过一个利用廉价、高度并行的GPU硬件的肽搜索引擎，大幅降低蛋白质组学实验的计算成本。因此，该项目如果成功，将使蛋白质组学更成功地用于公共卫生。研究团队完全有资格进行这项研究，他们在所有科学学科和必要的特定软件元素方面都有广泛的直接经验。研究团队包括蛋白质组学、质谱学、肽搜索和GPU计算方面的专家。

DESCRIPTION (provided by applicant): The use of Graphics Processing Unit (GPU) devices is proposed to increase the speed of peptide search by as much as an order of magnitude. Peptide search engines perform the most computation-intensive task in shotgun proteomics. This proposal is to speed up peptide search using the increasingly popular approach of general-purpose computation on highly-parallel GPU devices. Computation using GPUs to gain "desktop supercomputer" performance at low cost is an active and fruitful area of research. This project will obtain these benefits for peptide search. As a result, researchers will be able to take experiments that today must be analyzed on a computer cluster, and instead analyze them on their desktop computer. PUBLIC HEALTH RELEVANCE: A enhancement to the popular peptide search engine X!Tandem is proposed, to allow it to run in parallel on graphics processing unit (GPU) cards with dramatically faster performance. This will allow much faster and less-expensive operation, allowing proteomics researchers to analyze large proteomics experiments or search for post- translational modifications (PTMs) using their existing desktop computer. Proteomics has led to many important advances in biological understanding. Yet, many valuable data sets are not searched for PTMs, simply because the computer power necessary to conduct the searches is not available. With this project, we plan to substantially reduce the computational cost of proteomics experiments, via a peptide search engine making use of inexpensive, highly parallel GPU hardware. Thus, this project, if successful, will allow proteomics to be more successfully exploited for public health. The research team is well-qualified to undertake this research, having extensive, direct experience in all the scientific disciplines and specific software elements necessary. The research team includes experts in proteomics, mass spectrometry, peptide search, and GPU computing.