面向基因组重测序分析的混合异构体系结构关键技术研究

Genome re-sequencing analysis techniques, which is based on sequencing technology, are driving the technological revolution in the field of life science, and playing an important role in personalized medicine, rapid screening of single-gene diseases, large-scale study of the evolution of species, populations, public health emergency response and molecular breeding. It has become a fundamental task of bioinformatics and modern life science. With the development of high-throughput sequencing technology, the rapid expansion of the sequence data, massive data processing has become the serious challenges facing the field. High performance computer systems are widely used to accelerate sequence analysis applications. However, parallel efficiency is greatly limited by the diversification of program features: bit-wised parallelism, complicated and multi-dimensional data dependency, and tight synchronization resulting from irregular computing and storage features. Although general-purpose architecture computers have powerful peak performance, efficiently executing the task of genome re-sequencing analysis on a general-purpose parallel computer becomes very awkward. .In order to meet the timing needs of genome re-sequencing analysis, we must make full use of the heterogeneous computing system to increase computing speed of the bioinformatics software. There are several issues such as low computational efficiency and low resource utilization of this field. Our project is designed with full consideration of both algorithm optimization and architecture design. Based on the analysis of the algorithm design and optimization technology on computing platform with the CPU-GPU and the CPU-MIC general heterogeneous architecture, we would design and construct the FPGA-based reconfigurable accelerator. After that, a hybrid heterogeneous computing platform will be built with three types of accelerators, including GPU, MIC and FPGA. Different steps of the genome re-sequencing process will be mapped to different computing units according to the algorithm characteristics, and the whole process of genome re-sequencing analysis will be accelerated on our hybrid heterogeneous computing platform, so as to meet the need of timeliness of genome re-sequencing analysis.

基因组重测序分析是生物信息学乃至现代生命科学领域重要的基础性研究工作。随着高通量测序技术的发展，序列数据量急剧膨胀，海量数据处理已成为该领域面临的严峻挑战。此外，该领域的应用具有程序特征多样化、数据相关多维度、访存行为不规则等特点。通用结构计算机虽然能够提供很强的峰值计算能力，但不能完全适应该领域复杂计算特性的特殊要求，计算效率不高。.课题以基因组重测序分析典型处理流程的高时效性需求为背景，从并行算法设计和体系结构设计两方面出发，在深度研究基于CPU-GPU和CPU-MIC两类通用异构计算平台和基于FPGA的可重构计算平台的并行算法设计与优化技术的基础上，构建包含GPU、MIC和FPGA三种不同特性的算法加速器的混合异构计算平台，实现不同计算单元的体系结构特征与数据分析流程不同环节计算特征的适配，从而有效提高数据分析流程的整体计算性能，实现对基因组重测序分析全流程、系统化的加速解决方案。

课题以基因组重测序分析典型处理流程的高时效性需求为背景，从并行算法设计和体系结构设计两方面出发，在深度研究基于CPU-GPU和CPU-MIC两类通用异构计算平台和基于FPGA的可重构计算平台的并行算法设计与优化技术的基础上，构建包含GPU、MIC和FPGA三种不同特性的算法加速器的混合异构计算平台，实现不同计算单元的体系结构特征与数据分析流程不同环节计算特征的适配，从而有效提高数据分析流程的整体计算性能，实现对基因组重测序分析全流程、系统化的加速解决方案。. 本课题的研究工作主要包括以下几个方面：（1）基因组重测序分析算法的特征分析及性能模型，（2）基于通用异构体系结构的基因组重测序分析算法设计与优化技术，（3）基于可重构异构体系结构的基因组重测序分析算法设计与优化技术，（4）基于混合异构体系结构的原型系统构建，（5）基因组重测序分析流程可视化设计环境设计和构建。课题组成员目前已在国际期刊/会议发表论文14篇，其中发表期刊论文8篇（SCI检索6篇），国际会议论文6篇（EI检索6篇），另有4篇论文已投稿（SCI 检索2篇，EI 检索2篇）。申请国家发明专利5项，均已公示，申请软件著作权2项，已获授权。本项目研究工作已按期完成，主要研究成果及量化指标均达到预期目标。本项目研究有助于提高生物测序信息分析效率，并能为其它应用领域的算法并行化和提高典型应用的计算性能提供技术参考。