| Summary: | 碩士 === 國立嘉義大學 === 農學研究所 === 97 === Statistical methods for mapping quantitative trait loci (QTL) have been widely discussed. While most existing methods assume that the phenotype is normally distributed and completely observed, these assumptions may be violated when the phenotype is time-to-event, which has a skewed distribution and is often loss to follow-up. In this article, the effects of QTL on the censored failure-time phenotype was formulate through the semiparametric accelerated failure time model (Wei 1992). The approaches utilizing rank-based inference were developed to estimate regression parameters, and score tests, instead of traditional likelihood ratio tests, were employed to the entire chromosome for QTL. An extra challenge for mapping QTL is how to determine an appropriate threshold value for declaring the presence of QTL in multiple tests on the same genome. The resample approach proposed by Zou et al. (2004) was extended to determine the threshold value for multiple tests. Several simulation studies are presented to evaluate the performance of the proposed methods.
|
|---|
