XPS: FULL: DSD: Collaborative Research: Parallelizing and Accelerating Metagenomic Applications

XPS：完整：DSD：协作研究：并行化和加速宏基因组应用

• 批准号: 1533933
• 负责人: Yuan Xie
• 金额: $ 54.8万
• 依托单位: University of California-Santa Barbara
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-09-01 至 2020-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1533933&HistoricalAwards=false
• 关键词: XPS; FULL; DSD; Collaborative; Research

The importance of metagenomics arises from the fact that over 99% ofthe species yet to be discovered are resistant to cultivation. Unlikesingle genome sequencing, assembly of a metagenome is intractable andis in large part, an unsolved mystery. Moreover, the advent of highthroughput sequencing is fueling rapid generation of enormousmetagenomic datasets. There is no available sequenced genome for amajority of the species. There is a need to determine the number of species ina metagenomic dataset as well as the abundance of each of thesespecies. The key steps (Assembly and Clustering) in the metagenomicsanalysis algorithms are compute-intensive, while the sheer amount ofdata the algorithms operate on is staggering. The most promising wayto tackle the computational challenges is to build special purposehardware, dedicated solely to suitable algorithms.The main objective of this project is to develop a range of flexible,affordable, parallel, fast hardware-accelerated bioinformaticssolutions, using GPGPU, FPGA, and ASIC, for metagenomic analytics to provide an alternative toexpensive computer clusters. Specifically, hardware solutions formetagenomic clustering and assembly will be developed. Severalacceleration methodologies, including parallel software mapping andspecial hardware design, are proposed to explore the parallelisminside the applications and to improve the data access bandwidth inthe hardware running bioinformatics applications. The ideas proposedin this work will be evaluated in a multi-pronged manner using acombination of simulation, emulation and prototyping efforts. Further,the PIs will use a combination of commercial tools, collaboratorresources and existing internal tools. The research will be conducted in collaboration with industrial partners. Through closecollaboration with several industry partners, direct transfer of manyideas to industry is enabled. The outcome of this research will,therefore, have a direct impact on future bioinformatics applicationsolutions. This project will involve graduate and undergraduatestudents in all aspects of the research. The PIs will activelyintegrate the research results from this project into the graduate andundergraduate curricula, and develop new interdisciplinary courses onbioinformatics and computer architecture to train the next generationwork-force. Finally, the tools and techniques developed in thisresearch will be made available through web-sites for use by othereducators, researchers, and industry practitioners.

宏基因组学的重要性源于这样一个事实：超过 99% 的尚未发现的物种都对栽培有抵抗力。与单基因组测序不同，宏基因组的组装是棘手的，并且在很大程度上是一个未解之谜。 此外，高通量测序的出现正在推动巨大的宏基因组数据集的快速生成。大多数物种都没有可用的基因组测序结果。需要确定宏基因组数据集中的物种数量以及每个物种的丰度。 宏基因组分析算法中的关键步骤（组装和聚类）是计算密集型的，而算法操作的数据量是惊人的。 解决计算挑战最有希望的方法是构建专门用于合适算法的专用硬件。该项目的主要目标是开发一系列灵活、经济、并行、快速的硬件加速生物信息学解决方案，使用 GPGPU、FPGA 和 ASIC，用于宏基因组分析，为昂贵的计算机集群提供替代方案。具体来说，将开发宏基因组聚类和组装的硬件解决方案。提出了几种加速方法，包括并行软件映射和特殊硬件设计，以探索应用程序内部的并行性并提高运行生物信息学应用程序的硬件中的数据访问带宽。这项工作中提出的想法将结合模拟、仿真和原型设计工作以多管齐下的方式进行评估。此外，PI 将结合使用商业工具、合作者资源和现有的内部工具。该研究将与工业合作伙伴合作进行。通过与多个行业合作伙伴的密切合作，可以将许多想法直接转移到行业。因此，这项研究的结果将对未来的生物信息学应用解决方案产生直接影响。该项目将涉及研究生和本科生的各个方面的研究。 PI们将积极将该项目的研究成果融入到研究生和本科生课程中，并开发新的生物信息学和计算机体系结构跨学科课程，以培养下一代劳动力。最后，本研究开发的工具和技术将通过网站提供，供其他教育工作者、研究人员和行业从业者使用。

项目成果

MEDAL: Scalable DIMM based Near Data Processing Accelerator for DNA Seeding Algorithm

• DOI: 10.1145/3352460.3358329
• 发表时间: 2019-10
• 期刊: Proceedings of the 52nd Annual IEEE/ACM International Symposium on Microarchitecture
• 作者: Wenqin Huangfu;Xueqi Li;Shuangchen Li;Xing Hu;P. Gu;Yuan Xie