基于不完全双列杂交设计的水稻农艺性状配合力基因组预测

[目的]在亲本一般配合力的基础上优选特殊配合力高的杂交种,是水稻杂种育种的关键。基因组选择基于覆盖全基因组的分子标记和样本的表型数据建立预测模型,实现对品种更加可靠的选择。[方法]本研究利用一组基于不完全双列杂交(NCII 设计)的水稻数据集,考查其多个农艺性状配合力的基因组预测能力。并比较了不同训练群体构建方法对杂交种表型预测能力的影响。[结果]8个农艺性状一般配合力的预测能力由其遗传率主导,从0.3888 到0.7367。杂交种特殊配合力的预测能力较低,但是直接预测杂交种的表型可以获得较高的预测能力。[结论]基因组预测水稻亲本一般配合力是有效的,能够帮助育种家实现对亲本的科学选择。如果要选配杂交种,直接预测杂交种表型是最有效的手段。此时让更多的亲本均衡地参与杂交种训练集的组配,有利于获得更高的预测能力。

Genomic Prediction of Combining Ability for Agronomic Traits in Rice Based on NCII Design

【Objective】It plays a key role in hybrid rice breeding to select hybrids with high specific combining ability based on high general combining ability of parental inbred lines. Genomic selection that is based on molecular markers across the whole genome and phenotypes of samples enable us to establish prediction models and achieve more reliable selection of varieties. 【Method】We investigated the genomic predictive ability of combining ability for agronomic traits in rice based on NCII design. And the effects of different training population construction methods on predictive ability of hybrid performance were compared. 【Result】The predictive abilities of general combining ability for eight agronomic traits, ranged from 0.3888 to 0.7367, were dominated by their heritability. The predictive ability of specific combining ability for hybrids was lower, but the ability of directly predicting phenotypes for hybrids was higher. 【Conclusion】The genomic prediction of combining ability for rice parental lines is effective and can help breeders to select parents effectively. With regard to hybrid selection, direct predicting phenotypes of hybrids is the most effective method. At this time, allowing more parents to participate in the crosses for hybrid training set in a balanced way is benefical to obtain higher predictive ability.