Mapping Dynamic Trait Loci With Random Regression Models

Quantitative traits whose phenotypic values change with time or other quantitative factors are called dynamic quantitative traits. In order to describe the effects of QTL on changing process of dynamic trait, we nested Legendre polynomials with appropriate order into genetic and environmental effects in genetic model for mapping QTL based on the orthogonal, additive and universal-practicable properties of Legendre polynomials and constructed therefore the random regression models for mapping dynamic trait loci.With regard to random regression models mapping strategy for dynamic trait, we carried out following researches:1) Construction of random regression model with single and double dynamic trait QTL respectively, and summarizition of the expectation-maximization (EM) steps for parameter estimation based on the EM algorithm for a mixed model.2) Investigation of the mapping behaviors of the random regression models with single and double dynamic trait loci and comparison with functional mapping stategy by using Monte Carlo simulation.3) Application of random regression models mapping strategy for dynamic trait to mapping QTL responsible for the stem diameter growth of poplar trees on a linkage group. The results form simulations showed that in detecting dynamic traits loci with random regression model: (1) the power increases as the sample size increased, (2) the standard error of each estimated parameter decreases as the sample size increased, (3) the increased power and decreased error are also associated with the increased measurements frequency of dynamic traits. Both simulations and data analysis demonstrated that the random regression model with Legendre polynomial as submodels is more effective than the functional mapping with Logistic curve as submodel in mapping dynamic trait loci...