通过从头序列组装方法获得基因组序列是基因组学研究的基础。Scaffolding是从头序列组装中的关键步骤，它推断contigs之间的方向和顺序关系，能够使组装结果更加连续和完整，因此如何设计高效准确的scaffolding方法是需要解决的重要问题。本项目将通过挖掘和分析双端读数的统计特征对scaffolding方法展开研究。首先，基于同一contig相邻子区域中双端读数的多种特征变化，识别是否发生连接错误以及错误类型。其次，基于两个contigs之间相关双端读数的分布特征，提出一种更准确的contigs关联图构建方法。最后，通过对contigs关联图进行划分以减少后续的运算规模，并设计优化模型和算法确定contigs之间的方向和顺序关系。本项目的研究将为获取完整和准确的基因组序列，以及理解生命活动的内在组织和过程提供帮助。

Acquiring genome sequence by de novo sequence assembly tool is the foundation of genomics research. Scaffolding is the key step of de novo sequence assembly which infers the orientations and order of contigs, scaffolding is capable of producing more continuous and complete assembly results, how to develop efficient and correct scaffolding method is an important issue to be resolved. This project will be implemented based on mining and analyzing statistical characters of paired reads. Firstly, this project identifies the linking errors and categories based on the difference about characteristics of paired reads between two adjacent sub-regions in one contig. Secondly, based on the distribution of paired reads between two contigs, this project presents a more precise method to construct correlation graph of contigs. Finally, this project introduces one algorithm for partitioning correlation graph of contigs to reduce computing cost in the following steps, and designs optimization model and algorithm to determine the orientations and order of all contigs. This research is useful to get more complete and correct genome sequence, and will be helpful to understand the organization and process of life activities.