摘要
研究提出了一个同时含有主效基因和微效多基因的混合模型连锁分析方法。首先应用LASSO方法快速估计线性混合模型(LMM)非零的主效位点,并将这些位点效应视为固定效应;而后把基于约束极大似然法(REML)估计得到的微效多基因效应再放入LMM中作为随机效应;从而最终实现QTL遗传参数估计和基因定位连锁分析。针对系谱群体的定位分析,该模型,即LASSO-LMM法,不仅很好地继承了单独LASSO法的计算速度,而且还具有比单独LMM法更好的遗传参数估计稳健性。为了具体比较这3种定位方法,采用计算机模拟数据验证。结果表明,相比于单独LASSO法和单独LMM法,LASSO-LMM法在QTL遗传参数估计和统计效力上总体表现出更高的估计精确度和检测效率。
This study proposes a linkage analysis method of mixed model which contains both the major genes and the polygenes. Firstly, the compression technology of LASSO is used to quickly estimate the non- zero major gene loci in the linear mixed model (LMM) ; and then these loci effects regarded as fixed effects and the random polygenic genetic effects predicted by restricted maximum likelihood (REML) are simultaneously embedded into the LMM; finally we conduct the estimation of QTL genetic parameter and the linkage analysis of gene mapping. On account of the analysis of mapping in pedigree population, the model,so-called LASSO-LMM method, can not only inherit the computing speed of LASSO method, but also possess the better robustness of genetic parameter estimated than LMM method. In order to compare these three mapping methods in detail, we adopt the computer simulation datasets to make a validation. The result shows that compared with the other two methods, in terms of QTL genetic parameter estimation and statistical power, LASSO-LMM method totally performs a higher accuracy of estimation and detectableefficiency.
出处
《上海交通大学学报(农业科学版)》
2015年第4期65-70,76,共7页
Journal of Shanghai Jiaotong University(Agricultural Science)
基金
国家自然科学基金(30972077
31172190)