玉米萌发成苗期抗旱性全基因组预测及精度研究

The stage from germination to seedlings is the main period affected by the drought stress in maize production in the northern area of China. Traditional breeding methods show a low efficiency in the selection of maize drought-resistant varieties. Genomic selection (GS) would improve efficiency because it is one of the most suitable methods to predict complex quantitative traits controlled by multiple genes and, moreover, it is even not necessary to know QTL information or genetic basis in advance. However, lower prediction accuracy (rMG) limits its application in drought-resistant breeding. In this study, two populations, a bi-parental population, and an associated panel were used as materials for finding main factors affecting GS prediction accuracy. The different predicted results will be compared among the prediction algorithms, the training population sizes, the marker densities, the relationship between training populations and prediction populations, and the associated SNPs obtained by GWAS and QTL analysis to improve rMG as well as the key factor. The aim of this study is (1) to analyze the reason for the lower prediction accuracy of each method under the drought environment; (2) to optimize modeling to improve the prediction accuracy; (3) to explore more suitable prediction methods. Our research result would provide theoretical reference and technical support for drought-resistant breeding in Maize.

玉米萌发成苗期是北方玉米生产受干旱影响的主要阶段之一，通过传统育种方法选择抗旱材料效率不高，全基因组选择可弥补其不足。该技术不需要事先了解控制性状的QTL/基因，适合复杂的数量性状，但预测精度有待进一步提高。本项目拟以1个玉米双亲群体和1个自然群体为材料，通过室内和田间结合的方式获得萌发成苗期的抗旱表型数据，通过简化基因组测序获得基因组标记信息，利用不同算法建立预测模型，对玉米萌发成苗期抗旱性进行全基因组预测，研究不同算法、不同标记密度、建模群体大小、建模群体与被预测群体的关系和遗传力等对预测精度的影响，确定关键影响因子，同时对双亲群体和自然群体分别进行QTL和GWAS分析，并检测显效QTL/基因对全基因组预测的影响，探索建立新的更适合的预测方法，为改进抗旱性全基因组预测精度提供理论参考，并为全基因组技术在玉米抗旱育种中的应用提供技术支撑。

玉米是人类和畜牧不可或缺的全球作物，苗期遭受干旱会降低玉米的产量。全基因组预测已经在动植物中应用，并被视为加速遗传增益的有效工具，但是干旱胁迫下，复杂性状的预测精度有限。我们组建了378份遗传多样性丰富的自交系群体，并在DArT-seq和GBS平台下完成基因分型。我们评估了建模群体规模、标记密度、标记质量、显著标记对预测精度的影响。使用GWAS获得的显著标记做干旱预测时，平均预测精度得到了显著提高，提高幅度达30%-50%，建议在干旱条件下做全基因组选择前先利用GWAS选取显著标记。相比之下，而通过优化建模群体规模、标记密度、标记质量等因素，预测精度提高幅度较小。通过比较，DArT-seq利用不足一万个标记获得了比GBS约十万标记还高的预测精度，非常适合部署全基因组选择育种。室内实验的准确性高于田间，GWAS定位基因使用室内实验的数据。通过GWAS的FarmCPU方法，9个独立的与玉米苗期抗旱性状相关的显著单核苷酸多态性位点被发现，在每个独立显著SNP位点的上下游50kb区间内筛选出有可能与玉米幼苗抗旱性状有关的候选基因共41个，候选基因包括参与代谢途径基因、ATP、RNA结合途径基因、信号传导基因、转录调控途径基因、细胞转运途径基因和结构蛋白基因。我们的结果对玉米苗期干旱起到积极的作用，并有助于应对越来越频繁的极端天气对玉米产量的影响。此外，干旱条件下的杂交种预测、配合力预测、跨环境预测、病害预测等均有助于提高玉米品种的抗旱性，我们也做了相关研究，为进一步提高玉米抗旱育种打下良好的基础。