异质弦理论
杂种优势
最佳线性无偏预测
生物技术
生物
枯萎病
抗性(生态学)
育种计划
预测建模
基因组选择
训练集
数学
农学
单核苷酸多态性
基因型
遗传学
统计
计算机科学
人工智能
栽培
基因
选择(遗传算法)
混合的
数学物理
作者
Frank Technow,A. W. Burger,Albrecht E. Melchinger
标识
DOI:10.1534/g3.112.004630
摘要
Northern corn leaf blight (NCLB), a severe fungal disease causing yield losses worldwide, is most effectively controlled by resistant varieties. Genomic prediction could greatly aid resistance breeding efforts. However, the development of accurate prediction models requires large training sets of genotyped and phenotyped individuals. Maize hybrid breeding is based on distinct heterotic groups that maximize heterosis (the dent and flint groups in Central Europe). The resulting allocation of resources to parallel breeding programs challenges the establishment of sufficiently sized training sets within groups. Therefore, using training sets combining both heterotic groups might be a possibility of increasing training set sizes and thereby prediction accuracies. The objectives of our study were to assess the prospect of genomic prediction of NCLB resistance in maize and the benefit of a training set that combines two heterotic groups. Our data comprised 100 dent and 97 flint lines, phenotyped for NCLB resistance per se and genotyped with high-density single-nucleotide polymorphism marker data. A genomic BLUP model was used to predict genotypic values. Prediction accuracies reached a maximum of 0.706 (dent) and 0.690 (flint), and there was a strong positive response to increases in training set size. The use of combined training sets led to significantly greater prediction accuracies for both heterotic groups. Our results encourage the application of genomic prediction in NCLB-resistance breeding programs and the use of combined training sets.
科研通智能强力驱动
Strongly Powered by AbleSci AI