分类性状基因定位的多性状似然分析法  

Mapping QTL for Categorical Trait Using Multiple Trait Maximum Likelihood Analysis

作　　者：田佺[1] 杨润清[1] 

机构地区：[1]上海交通大学农业与生物学院,上海201101

出　　处：《上海交通大学学报（农业科学版）》2005年第2期111-116,共6页Journal of Shanghai Jiaotong University(Agricultural Science)

基　　金：上海交通大学博士基金资助(A2737A)

摘　　要：简单线性模型统计方法是业已被证明的与GLM法具有相近参数估计准确度和检测效率的一种二歧性状基因定位分析的简化方法。将适于二歧性状的一元线性模型推广到多类别分类性状基因定位分析中,文中提出采用多元线性模型对多类别分类性状进行基因定位分析。该方法首先将分类性状的各个类别转化为不同的二歧性状,并把每个二维数据看作是连续的,然后采用似然分析法,根据多性状联合定位的统计准则判断QTL的位置和估计QTL效应。模拟结果表明:新方法在大样本和中等遗传力条件下,能够以很高的计算效率相当准确地估计分类性状隐含变量给定的QTL位置和效应。It has been validated that the simple linear model approach, whic h has similar results with generalized linear model approach, is a simplified meth od for mapping QTL of binary trait. We developed a multiple trait linear model a pproach to mapping QTL for categorical trait from simple linear model approach t o mapping QTL for binary trait. This approach treated each category in categoric al trait as binary trait and constructed multiple trait linear model by treating each binary trait as continuous, and then adopted maximum likelihood analysis t o estimate QTL position and effects based on the test statistic for jointly mapp ing QTL of multiple traits. The method proposed here is not only able to detect QTL controlled whole categorical trait, but also responsible for each category i n categorical trait. Its validity was demonstrated by a simulation.

关 键 词：分类性状 分析法 多性状 基因定位 定位分析 一元线性模型 多元线性模型 QTL效应 检测效率 参数估计 统计方法 二维数据 统计准则 联合定位 模拟结果 隐含变量 计算效率 准确度 类别 遗传力 大样本 

分 类 号：O212.1[理学—概率论与数理统计] Q949.747.1[理学—数学]