全基因组关联荟萃分析的新方法研究与应用

Genome-wide association study (GWAS) analysis has become an important tool for identifying the genetic basis of complex traits in plants. Genome-wide association meta-analysis (MetaGWAS) is a recently proposed methodology that improves the accuracy of detecting trait-related genetic variants through combining GWAS data from different studies. Although several mathematical and computational models have been presented for MetaGWAS, all were developed based on human GWAS data, the performance of which on plant GWAS data is still unknown. Moreover, they are based on traditional statistical methods (e.g., Fisher's method) that are difficult to correctly estimate the distribution of model parameters. and to effectively integrate multiple functional genomic datasets. Additionally, no bioinformatics platform with multiple functional modules has been developed for MetaGWAS analysis. In this proposal, we will first systematically evaluate the performance of existing MetaGWAS methods using simulated and real plant GWAS data, and pinpoints the strengths and weaknesses of each method, and develop a computational pipeline specifically for plant MetaGWAS analysis. Then, we will introduce machine learning technologies into MetaGWAS analysis, and develop a new computational model with higher prediction performance than existing method in identifying trait-related genetic variants, by integrating GWAS analysis results from individual studies with large-scale functional genomics data. Subsequently, we will develop a bioinformatics platform with a user-friendly interface for GWAS data management, meta-analysis and visualization, without requiring any local installation and programming experience. Finally, we will perform a biological application of constructed bioinformatics platform on maize GWAS data consisting of a collection of multiple agriculturally important traits. The implement of this project will facilitate the development of new methodologies, new software and bioinformatics platforms in plant GWAS research fields, and will accelerate the dissection of genetic mechanisms of complex traits in maize and other crops.

全基因组荟萃分析（MetaGWAS）是综合多组独立的全基因组关联研究(GWAS)，可更准确地检测性状相关遗传变异的GWAS分析新方法。MetaGWAS还存在如下问题：依据人类GWAS数据研制，在植物GWAS数据上的性能尚不清楚；基于传统统计学方法构建的数学模型需要预先假定模型参数分布，且难以整合功能基因组学数据；缺乏一个功能集成的MetaGWAS生物信息学分析平台。本研究拟对现有MetaGWAS分析方法在植物GWAS数据上展开系统的测评，发现其优缺点，建立一套适合植物的MetaGWAS分析流程；利用人工智能中的机器学习方法整合多种功能基因组学数据，建立高精度而且快速的MetaGWAS分析新方法；构建一个集GWAS数据管理、荟萃分析和结果可视化于一体的生物信息学分析平台，并在玉米GWAS数据上展开应用研究。本项目的实施将有助于加快玉米等农作物复杂性状分子遗传机制的解析进程。

全基因组荟萃分析（MetaGWAS）是综合多组独立的全基因组关联研究(GWAS)，可更准确地检测性状相关遗传变异的GWAS分析新方法。MetaGWAS还存在如下问题：依据人类GWAS数据研制，在植物GWAS数据上的性能尚不清楚；基于传统统计学方法构建的数学模型需要预先假定模型参数分布，且难以整合功能基因组学数据；缺乏一个功能集成的MetaGWAS生物信息学分析平台。本项目测评了代表性的MetaGWAS方法在植物GWAS数据上的性能，研发了基于机器学习技术的高通量测序数据充分利用、基因型到表型精确映射、植物重要功能挖掘的新方法，建立了一个集GWAS和荟萃分析于一体的植物MetaGWAS的生物信息学分析平台。本项目的研究成果将有助于深入挖掘植物功能基因组的价值，加快植物复杂性状分子遗传机制的解析进程。