解析农作物产量性状相关的遗传构成和调控机制是提高粮食单产的重要基础，本项目从生物信息学角度，针对影响主要性状的关键基因发现问题，发展全基因组关联分析（GWAS）的有效算法，并研究面向多数据融合的遗传调控网络模型。首先，采用聚类方法，并根据连锁不均衡特性，建立基于图模型的tagSNP获取算法，为GWAS研究提供规模合适的代表性SNP位点。然后，针对GWAS在全基因组范围内检测上位效应计算量大的问题，提出基于遗传算法的多SNP位点上位效应检测方法；同时，拟提出基于多阈值的先验知识表示方法，利用SNP先验信息提高算法检测性能。最后，拟给出面向多种组学数据融合的动态贝叶斯网络结构评分函数和学习算法，为遗传调控网络解析提供合理的模型描述。本项目的研究有利于人们深入理解粮食作物复杂性状的遗传构成和调控机制，为我国农作物的高产育种研究提供新的信息学方法，因此，本项目具有重要的研究意义和潜在的应用价值。

It is the important basis for improving grain yield to analyze the genetic structure and regulation mechanism related to crop yield. Aiming at the key genes mining problems that affect the main characters, the project tries to develop the effective algorithms of genome-wide association studies (GWAS), and to study the genetic regulatory network model for data fusion from the point of view of bioinformatics. First of all, with the clustering method and the linkage disequilibrium (LD) characteristics, the tagSNP selecting algorithm is proposed based on graph model to provide the representative SNP site scale suitable for GWAS research. Then, for the epistasis detecting problem with the high complexity on a genome-wide scale in GWAS, we will study the detection method based on genetic algorithm for multiple SNP sites epistasis; and in the meantime, we propose a SNP domain knowledge representation method based on multi-threshold, to improve the detection performance of algorithm. Finally, the project tries to present the score function and learning algorithm of dynamic Bayesian network (DBN) structure to integrate multiple omics data so as to provide the reasonable model for genetic regulatory network analysis. The research is helpful for people to deeply understand the grain crop complex genetic structure and regulation mechanism of traits, and provide new information science method for the study of high yield breeding in China agriculture. Therefore, this project has important research significance and potential application value.