家畜遗传抗性多基因效应预测的广义线性方法  

Methodology of Predicting Additive Polygene Effect for Hereditary Resistance Using Generalized Linear Method

作　　者：殷宗俊[1] 张勤[2] 张纪刚[2] 刘剑锋[2] 

机构地区：[1]安徽农业大学畜牧水产学院,合肥230036 [2]中国农业大学动物科技学院,北京100094

出　　处：《畜牧兽医学报》2006年第4期313-316,共4页ACTA VETERINARIA ET ZOOTECHNICA SINICA

基　　金：国家重点基础研究发展规划(G2000016103);安徽省自然科学基金(050410204);安徽省教育厅项目(2002jq126;2004kj151)

摘　　要：在广义线性混合模型(GLMM)的框架内模拟研究了抗性性状的遗传分析方法,初步比较了GLMM方法与一般线性方法(LM)的育种值估计效果。模拟研究的抗性表型为单阈值和3阈值2种类型,选用的连接函数分别为logist连接和log连接,试验设计为全同胞-半同胞混合家系,参数估计采用Fisher迹法。研究结果表明,GLMM方法能较准确地估计公畜的个体育种值,在个体的遗传评定效果方面要明显优于LM方法,其预测的个体育种值排序结果与真实育种值的排序之间存在显著的秩相关。The objective of this paper is to provide an introduction to generalized linear mixed models and to compare of the efficiency of the GLMM and LM. The method of GLMM is described that allows genetic analyses and predicting breeding value for resistance traits. Binary and ordinal response traits are simulated in this study, and the design is mixed family of full-half sib. The logit and log linkage function are applied, respectively. The method of Fisher score is applied to calculate the parameters in generalized linear mixed models. The results showed that generalized linear mixed model can predict breeding value to a nicety. In addition, the method of GLMM has a great advantage in predicting the breeding values for resistance traits.

关 键 词：广义线性混合模型(GLMM) 遗传抗性 估计育种值(EBV) 

分 类 号：S813.1[农业科学—畜牧学]