BIGDATA: Mid-Scale: DA: Collaborative Research: Genomes Galore - Core Techniques, Libraries, and Domain Specific Languages for High-Throughput DNA Sequencing

最近出现的各种高通量DNA测序仪器，以及随之而来的每碱基成本的快速下降，正在生命科学的所有领域造成严重的数据泛滥。测序仪器的异质性和它们所支持的应用程序的巨大多样性正在为生物信息学社区创造许多分析问题。此外，传统的串行算法一直是生物信息学研究的支柱受到日益增长的数据集的严峻挑战。拟议项目的目标是开发核心技术和软件库，为高通量DNA测序（也称为下一代测序（NGS））提供可扩展、高效、高性能的计算解决方案。为了使更大的社区，该项目寻求1）确定一组经常出现在许多类型的高通量测序应用中的核心功能，2）为其开发高效的并行算法和高性能实现，3）追求到HPC架构的映射，包括集群，多核和GPU，4）开发封装这些功能的软件库，目标是使生物信息学社区能够利用HPC架构，以及5）设计领域特定语言，以使不熟悉并行处理的生物信息学研究人员能够通过自动生成并行代码从这项工作中受益。该研究将在人类遗传学和宏基因组学的挑战性问题的背景下进行，与领域专家合作。该项目的重点是一个关键的能力建设活动，以促进NGS生物信息学研究人员和从业人员普遍使用并行性。其目标是使更广泛的社区能够从智能并行算法、高度优化的实现和专用HPC硬件中受益，而不需要任何这些方面的专业知识。软件库将作为开源发布，供社区使用、进一步开发、增强和合并。该项目将为培养大数据分析和计算机科学驱动的跨学科研究的博士后和研究生提供机会。将利用伙伴机构现有的各种机制，推进少数民族和妇女招聘、本科生参与研究和K-12外联的目标。

The recent emergence of a variety of high-throughput DNA sequencing instrumentation, and the concomitant rapid decline in the cost per base, is causing severe data deluge in all areas of life sciences. The heterogeneity of sequencing instrumentation and the vast diversity of applications enabled by them are creating numerous analytics problems for the bioinformatics community to address. In addition, the conventional serial algorithms that have been the mainstay of bioinformatics research are severely challenged by the ever increasing data sets. The goal of the proposed project is to develop core techniques and software libraries to enable scalable, efficient, high performance computing solutions for high-throughput DNA sequencing, also known as next-generation sequencing (NGS). To empower the larger community, the project seeks to 1) identify a set of core functionalities that frequently occur in many types of high-throughput sequencing applications, 2) develop efficient parallel algorithms and high performance implementations for them, 3) pursue mapping to HPC architectures including clusters, multicores, and GPUs, 4) develop software libraries encapsulating these functionalities with the goal of enabling the bioinformatics community to exploit HPC architectures, and 5) design a domain specific language to enable bioinformatics researchers unfamiliar with parallel processing to benefit from this work through automatic generation of parallel codes. The research will be conducted in the context of challenging problems in human genetics and metagenomics, in collaboration with domain specialists.This project is focused on a key capacity building activity to facilitate pervasive use of parallelism by NGS bioinformatics researchers and practitioners. The goal is to empower the broader community to benefit from clever parallel algorithms, highly tuned implementations, and specialized HPC hardware, without requiring expertise in any of these. The software libraries will be released as open source for use, further development, enhancements, and incorporation by the community. The project will provide opportunities for training postdoctoral and graduate students in bigdata analytics and computer science driven interdisciplinary research. Diverse existing mechanisms at the partner institutions will be leveraged to advance goals of minority and women recruitment, undergraduate participation in research, and K-12 outreach.