基于Nanopore测序和多源数据融合策略的基因组大型结构变异检测方法研究

Structural variations (SVs) are genetic variations that occur on DNA. Studies have shown that SVs not only determine the differences in phenotypes of different individuals, but also are closely related to the generation of complex diseases such as cancer, schizophrenia, dementia and autism, and optimization of crop traits, breading and increased yield. Interpreting SVs in individual genomes can provide accurate and targeted solutions for research and diagnosis of complex diseases, optimization of plant properties, research and development of new drugs, and individual health management. Existing SVs detection methods are limited by sequencing technologies, and their detection coverage, accuracy and sensitivity are limited, causing the detection results mainly consist of the small mutations with the high false positive rates. Therefore, the role of large SVs in the genome in genetics and mutation has not been fully and accurately revealed. From the perspective of multi-source data fusion, this project is based on Nanopore sequencing data to study the efficient identification method for large SVs in genome, combining the technical characteristics of different sequencing steps of Nanopore, the multi-dimensional features of the low-quality subsequences, and the alignment characteristics between multi-source sequencing data. This study will provide support for revealing the role of large SVs in genetics and mutation.

结构变异(SVs)是发生在DNA上的遗传变异。研究表明，SVs不但决定着不同个体表型的差异，还与癌症、精神分裂症、老年痴呆症、自闭症等复杂疾病的产生，农作物性状优化、育种、增产等息息相关。解读个体基因组中的SVs可为复杂疾病的研究与诊断，植物性状优化，新药研发，以及个体健康管理等提供精准、针对性强的解决方案。现有SVs检测方法受测序技术的限制（二代测序片段跨度小，三代测序片段错误率高），其检测覆盖范围、精度、灵敏度都有限，检测结果以小型变异为主且假阳率高。因此，基因组中大型SVs在遗传和变异中发挥的作用还未被全面、准确的揭示。本项目将以Nanopore测序数据为基础，从多源数据融合的角度出发，基于Nanopore不同测序环节的技术特点、低质量子序列的多维特征，以及多源测序数据之间的比对特性研究基因组中大型SVs的高效识别方法。本研究将为揭示大型SVs在遗传和变异中发挥的作用提供支持。

重复序列在动物、植物、细菌等基因组中普遍存在。例如，人类基因组中重复序列的占比约为50%左右。大量研究表明，基因组中的重复序列在生物体的遗传、进化、变异、基因表达、转录调控、染色体重组和生理代谢等过程中起着不可或缺的作用，并与癌症等复杂疾病的发生有着密切的关联。重复序列的全面检测和精准注释是研究其在生物体生命活动中发挥的作用和探索其与复杂疾病之间关联的重要基础。然而，由于测序技术的限制和检测模式设计存在的缺陷，现有的检测方法在检测规模和精度方面都无法达到令人满意的程度，这严重制约了下游分析和应用的开展。为了突破测序技术的限制，克服现有检测方法在检测模式上的不足，项目主持人在充分分析重复序列的比对特性及其对序列组装过程的影响的基础上，基于多比对unique k-mers和全局性组装策略提出了一种全新的重复序列识别框架LongRepMarker和多物种重复序列数据库msRepDB。综合测评表明，与现有的检测方法相比, LongRepMarker在检测规模和检测精度上都具有明显优势，msRepDB与现有的Dfam和RepBase 数据库相比在收集物种数目和单一物种重复序列的精确度、完整性方面存在优势。