高精度和快速计算F2群体主效与环境互作QTL检测的理论与应用研究

F2 population contains additive and dominant effects, which can be used to study the genetic basis of heterosis and important traits in order to guide crop breeding. Although more and more linkage studies in F2 have been reported in recent years, it is harder to detect small effects and linked QTL, and also it is not easy to detect environment interaction. The possible reason is that the polygenic background control is not ideal and the single-locus models are only used. Up to now, quantitative traits are controlled by a few number of genes with large effects and a large number of genes with small effects, and also there is a clustering phenomenon. Therefore, the detection of small effects and linked QTL is crucial. Here, the applicant plans 1) to extend her FASTmrEMMA algorithm in GWAS, which can fast and efficiently estimate three variance components, in order to apply the polygenic background control in GWAS to QTL location, to estimate more than five variance components including additive and dominant polygenic background, and to implement multi-locus analysis with the whole genome scan for all the potential QTL, and to set up a new, high power, high accuracy and fast method for small and linked QTL; 2) the new method based on 1) is further extended to the detection of QTL and environment interaction; 3) A series of Monte Carlo simulation experiments and real data analysis are used to validate the new method; 4)Once the new method is confirmed, we will develop a software package with Windows interface. Two SCI papers need to be published, and a software package needs to be developed.

F2群体包含了加性与显性效应，是研究杂种优势和重要性状的遗传基础，以指导作物育种。虽然其QTL定位方法较多，但是小效应和连锁QTL检测较为困难，其环境互作更不容易。其可能原因是多基因背景控制不理想和只用单位点模型。业已表明：数量性状由少数大效应和多数小效应QTL控制，且有成簇现象。因此，检测小效应和连锁QTL十分重要。为此，本项目1)将关联分析多基因背景控制思想用于QTL定位，拓展申请人已提出的只能估计3个方差组分的高效快速全基因组关联分析新算法FASTmrEMMA，用来估计包含加性与显性多基因背景方差的不少于5个方差组分，通过全基因组扫描找出所有潜在QTL以实施多位点分析，建立高功效高精度快速计算的检测小效应和连锁QTL新方法；2)将1)的方法拓展至QTL与环境互作检测；3)通过模拟数据和实际数据分析，证实新方法的有效性；4)研制可供应用者使用的软件包。发表SCI论文2篇和研制软件1套。

F2群体是研究杂种优势和重要性状的遗传基础。但当前的方法检测F2群体小效应和紧密连锁QTLs比较困难，导致大多数数量性状和组学相关性状的遗传基础尚不十分清楚。为此，本项目完成了三部分工作：(1)提出了F2群体小效应和连锁QTLs检测的GCIM算法，它包括GCIM-random和GCIM-fixed。利用GWAS中估计多基因方差代替连锁分析中选择相关标记控制多基因背景，分别对F2群体加性效应和显性效应实施FASTmrEMMA算法，选取P-值负对数曲线峰值位点作为潜在SNPs，放入到多位点模型中，利用自适应LASSO算法及似然比检验检测与性状显著相关的QTLs。这就是GCIM-random，它能快速估计5个方差分量。当加性和显性效应被视为固定效应时，这称之为GCIM-fixed。模拟研究和真实数据分析表明，GCIM算法可以显著提高F2群体小效应和连锁QTLs的统计功效，检测到更多的小效应基因。(2)提出了大群体的FASTmrEMMA和GCIM加速算法。利用Woodbury等矩阵恒等式和性质，将FASTmrEMMA算法单标记扫描的特征向量计算转换为向量内积运算和向量-矩阵-向量乘法运算，计算速度比原算法提高了至少60%。将FASTmrEMMA加速算法拓展到GCIM-random，即得到适用于F2大群体的GCIM加速算法，其计算速度提高了至少20%。(3)开发了R软件包QTL.gCIMapping v2.0，它能够对BC、DH、RIL和F2等双亲群体进行QTL作图；开发了R软件包mrMLM v4.0，它包括了FASTmrEMMA加速算法等6种多位点全基因组关联分析算法。此外，利用FASTmrEMMA算法的矩阵变换，提出了基于最小角回归和随机森林的GWAS两步算法TSLRF。发表SCI论文5篇，其中高水平论文2篇。取得软件著作权1项。