摘要
综合考虑主基因效应以及基因间的交互效应对植物选育种的作用是基因组选择研究关注的热点问题之一.目前已有的研究大多忽略了基因的交互效应,这主要是由于考虑交互效应会大大增加备选基因的数目,从而导致已有的统计建模方法不稳定.本文将基因效应与基因间的交互效应同时引入模型,提出三步模型构建方法以达到简化计算和提高模型预测精度的目标.第一步,不考虑具体模型,通过距离相关筛除方法删掉与响应变量显著无关的基因;第二步,在剩下的基因中,利用贝叶斯方法筛选可能的基因;第三步,基于选出的基因,同时考虑单基因效应和交互效应,利用惩罚方法选择模型并估计参数.通过模拟计算说明我们提出的方法与已有的一步模型选择方法相比具有计算简单、稳健、运行时间少并且预测精度高等优点.最后,将本文的方法应用于油菜花数据,实证分析表明,我们提出的方法显著地提高花期性状的预测精度.
It is one of the hot topics in genome selection research to comprehensively consider the effects of main gene and intergenic interaction on plant breeding.At present,most existing studies ignore the interaction effect of genes,which is mainly because the interaction effect will greatly increase the number of candidate genes,resulting in the instability of existing statistical modeling methods.In this paper,gene effect and interaction effect between genes are introduced into the model at the same time,and a three-step model construction method is proposed to simplify the calculation and improve the prediction accuracy of the model.In the first step,genes significantly unrelated to response variables were deleted by distance correlation screening without considering the specific model;in the second st ep,the remaining genes were screened for possible genes by Bayes method;the third step is to select the model and estimate the parameters based on the selected genes and considering the single gene effect and interaction effect.Compared with the existing one-step model selection method,the proposed method has the advantages of simple calculation,robustness,low running time and high prediction accuracy.Finally,the method of this paper is applied to the data of rape flower,and the empirical analysis shows that the proposed method can significantly improve the prediction accuracy of flowering traits.
作者
刘妍岩
王蕊
赵燕
邹君
LIU YANYAN;WANG RUI;ZHAO YAN;ZOU JUN(School of Mathematics and Statistics,Wuhan University,Wuhan 430072,China};School of Sciences,Henan University of Technology,Zhengzhou 450001,China;College of Plant Science and Technology,Huazhong Agricultural University,Wuhan 430070,China)
出处
《应用数学学报》
CSCD
北大核心
2019年第5期684-700,共17页
Acta Mathematicae Applicatae Sinica
基金
国家自然科学自然基金(No.11571263)
国家重点研发计划(No.2016YFD0101300和2017YFC1600601)资助项目
关键词
交互效应
变量筛选
贝叶斯方法
interaction effect
variable selection
Bayes method