品种资源群体抗性性状QTL互作检测新方法及其应用

The tolerance to abiotic stress and resistance to diseases and insect pests in crop cultivar resource is not inherited in a simple Mendelian fashion. Their phenotypic observations are in discrete form and fellow non-normal distribution, indicating less information available and complex biological pattern. All these characteristics make it more challenging to dissect the genetic basis of resistance traits in crop cultivars. It should be noted that resistance traits are important targets in crop breeding, and have serious effects on crop production and quality. As we know, breeding new cultivar by making use of elite alleles mined from crop germplasm resource is the most economic and efficient way to reduce agricultural production loss. However, relatively little is known about statistic method of mining elite genes in crop germplasm resource, and especially for the existence of epistasis between genes. Based on epistatic association mapping of quantitative trait and multi-QTL mapping of resistance traits in crop cultivars, therefore, in this project the currently ideas from Na?ve Bayesian analysis, quasi-likelihood function method, hierarchical generalized linear model, Bayesian shrinkage estimation approach, penalized maximum likelihood and LASSO are used to develop a new algorithm of parameter estimation for epistatic genome-wide association studies of resistance traits in crop cultivars. Its purpose is to construct a technologic plarform for detecting epistasis of binary and ordinal traits (resistance traits) in crop cultivars. Monte Carlo simulation experiments are used to validate the new approach. The validated method is used to detect epistatic QTL of alkaline-salt tolerance to soybean. Two SCI papers will be published in this project..

作物品种资源群体对逆境和病虫害的耐性或抗性呈非孟德尔遗传的间断型非正态分布，其信息量少，内在生物模式复杂，使抗性性状的遗传剖析更具挑战性。抗性性状是重要的作物育种目标，影响作物产量和品质。利用品种资源群体发掘的优异基因进行新品种选育是减少农业生产损失最经济有效的办法。然而，在品种资源群体中检测抗性性状QTL的统计方法研究还相对薄弱，特别是十分重要的上位性互作。为此，在品种资源群体数量性状上位性关联分析方法和抗性性状多QTL检测方法前期工作基础上，利用朴素贝叶斯方法、伪似然函数法、分层广义线性模型、贝叶斯方法、惩罚最大似然法和LASSO等数理统计方法的思想，提出品种群体抗性性状上位性互作检测的参数估计算法；构建品种群体二歧和多歧抗性性状的上位性互作检测的技术平台；经Monte Carlo模拟研究验证后，用于大豆对盐碱耐性的上位性互作检测，并进行分子育种设计。拟发表SCI论文2篇。

抗性性状是重要的作物育种目标，但它不服从孟德尔遗传且信息量少，使得抗性性状遗传剖析更具挑战性，特别是其上位性互作检测方法。为此，本项目完成了三部分工作。首先，基于分层广义线性模型、贝叶斯方法、最大似然法，提出了品种资源群体二歧和多歧抗性性状上位性关联分析方法，可同时检测主效QTL、上位性和QTL-环境互作，并由一系列MonteCarlo模拟研究验证了方法的有效性。考虑到模型超饱和问题，提出用卡方测验对大量SNP标记进行初筛，然后再由该上位性关联分析方法进行检测。通过模拟实验确定阈值为0.25。利用本项目新方法对大豆幼苗根长盐碱耐性指数分类数据进行了上位性互作关联分析，共检测到12个QTL。其次，获得了2014和2015年的286个品种的大豆幼苗5个性状的耐盐碱性数据，经分析该品种群体存在显著遗传变异。最后，获得286个大豆品种群体的SNP数据，并已用于大豆开花时间与种子大小相关的驯化位点分析。此外，提出的mrMLM方法可通过多位点混合线性模型有效提高全基因组关联分析的功效与精度；并用数量性状上位性关联分析方法检测了该大豆品种群体基因间互作关系，剖析数量性状的遗传基础。发表3篇SCI论文和2篇中文核心。