| Summary: | 碩士 === 國立嘉義大學 === 農學研究所 === 97 === Most of approaches for mapping quantitative trait loci (QTL) require a well-known phenotype distribution ,e.g., normal, log-normal, or Weibull. These approaches may not be applicable when the phenotype distribution is doubtful and unspecified In this paper, using the semiparametric accelerated failure time model (Wei 1992) for the effect of QTL on the censored failuer-time phenotype, we developed approaches utilizing rank-based inference and empirical likelihood in estimating regression parameters and searching the entire chromosome for QTL. Some simulation studies were conducted to evaluate the performance of the proposed methods. Results from the simulated data indicated that both the proposed estimators of the QTL location and the QTL effects had only a little bias, and the efficiency would increase as the numbers of marks increased. In addition, the proposed semiparametric modeling strategy would obtain desirable robustness properties when compared with the parametric modeling strategy.
|
|---|
