面向海量原始多组学数据的序列快速检索方法研究

随着下一代基因组测序技术的发展，组学原始测序数据成爆炸式增长。国际上，NCBI的SRA库存储的序列长度已超过50Pb，其中可公开访问的序列超过20Pb。国内，国家基因组科学数据中心的组学原始数据归档库GSA所存储的数据量也超过了7PB。随着基因测序数据的快速增长，传统的基于序列比对的检索算法已经不适用于大规模数据检索。近年来，基于手绘草图（sketch）的算法和技术的研究，在计算生物学领域广泛开展，主要应用于序列聚类、物种分类等研究中。本课题将面向GSA数据库中的海量原始多组学数据，基于手绘草图的算法思想，研发一种基于布隆过滤器（Bloom Filter）的K-mer索引库，并在此基础上实现基因序列快速检索工具。该工具将使用GSA中的公开数据建立索引库，实现面向海量原始多组学测序数据的快速序列查询，进而可以应用于病原微生物溯源、临床微生物学研究以及公共卫生研究中。

Exponentially increasing amounts of raw sequencing data are stored in the global archives. The ability to query these data for sequence search-terms would facilitate both basic research and applications such as for pathogenic microorganism traceability, and clinical microbiology research. Traditional methods, usually based on sequence alignment methods, are not capable of handle these retrievals. In recent years, probabilistic methods such as sketching are introduced to computational biology to solve the problem. In this research, we will develop a new searchable K-mer tool to solve this problem. We will use the Bloom Filter as a sketching structure, then compress the K-mers decomposing from the GSA dataset to a condensed K-mer index database. Fast search of the GSA raw sequence data can be realized based on the index database. The tool will be used in pathogenic microorganism traceability, clinical microbiology researches and public health researches.