Regression Analysis Approaches to Combine the QTL Information in Multiple Quantitative Traits Using Sib-pairs Data

• Main Authors: Chao-Hsien Lee, 李昭憲
• Other Authors: 戴政
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/48453938044854981863

博士 === 國立臺灣大學 === 流行病學研究所 === 92 === Gene mapping for quantitative trait loci (QTL) was a difficult assignment in modern genetics. It was necessary to utilize the specific genetic design and strategy and the correctly statistical method for separating the genetic factors on the variations of quantitative trait values because quantitative traits were simultaneously affected by the genetic effect and environmental effect. The methods for analyzing QTL were established by the models that were inclusive of the explanations of traits. Different methods were modeled for different data structures which were gained from different genetic design. In this study, we tried to combine the QTL information in multiple quantitative traits using sib-pairs data. We considered and concerned that if the QTL influenced more than one trait and those influenced traits were individually analyzed for detecting genetic linkage; the problems about multiple comparisons were certainly occurred and arose. To avoid the problems of multiple comparisons and to increase the power for testing genetic linkage, we developed the method which could analyze simultaneously the influenced traits for detecting genetic linkage in two stages. In the first stage, we estimated the regression coefficients contained the information of genetic linkage by utilizing multiple regression model and polytomous logistic regression model. To consider the QTL information in multiple quantitative traits, we set the estimate of IBD as the dependent variable and the influenced traits as the independent variables in these models. In the second stage, we performed the method of combining tests for the regression coefficients obtained in the first stage to detect genetic linkage. Finally, we simulated the sib-pairs data to illustrate the performances of the power and the control of type I error in our method.