Methodology For Multiple Quantitative Trait Loci Mapping In An F_2:3 Design

Due to low heritabilities for most important economic traits in animal and plant sciences, the precision of the inheritance analysis for the above traits is relatively low. In order to improve the accuracy, geneticists have been exploring the methodology. In the real data analysis of quantitative trait loci (QTL) mapping in corn, cotton and cucumber, for example, an F_2:3 design, which is genotyped in F₂ plants and phenotyped in the F_2:3 progeny, is widely used. However, the methodologies for the F_2:3 design do not consider the mixture distribution for heterozygous F₂.3 progeny. Zhang & Xu (2004) indicates that this neglect will significantly reduce the power of QTL detection. Recently, this result is further confirmed by Hu & Xu (2006) from endosperm QTL mapping and Zhu et al. (2007) from resistance trait loci mapping. Therefore, it is essential to consider the above mixture distribution. At the present, the methodologies for the F_2:3 design is mainly based on a single-QTL model. Although Zhang & Xu (2004) proposed Bayesian analysis of multi-QTL mapping, detailed simulation studies and real data analysis were not presented. In addition, up to date the epistatic detection in the above design has been not reported.Many researches show that multi-QTL mapping is better than single-QTL mapping and epistasis plays an important role in the genetic architecture of quantitative traits. Therefore, the purpose of this dissertation is to develop the methodologies for multi-marker analysis, multi-QTL analsysis and epistatic QTL detection in the F_2:3 design. These methods were verified by a series of Monte Carlo simulations. The results were as follows.1) Multi-marker Bayesian analysis in the F₂.3 design. Because that to consider the mixture distribution and to adopt multi-QTL analysis in the F_2:3 design may increase the power of QTL detection, a Bayesian analysis method simultaneously using all markers on the entire genome was proposed in this dissertation. Monte Carlo simulation experiments show that the new method is better than the old-adopted F₂ method and interval mapping, and the power of QTL detection and the precision of the estimates for the positions and the effects of QTL increase as the number of F_2:3 families or the number of plants of each F_2:3 family increases. In addition, two sampling strategies for QTL effect were compared, and the new strategy that sampling value is conditionally accepted is better than the old one.2) Multi-QTL Bayesian analysis in the F_2:3 design. As stated above, the positions of QTL in the multi-marker Bayesian analysis were not considered. This neglect would bias the estimates for the positions and the effects of detected QTL away from linked markers. Therefore, it is necessary to propose multi-QTL Bayesian analysis approach. The results from simulation experiments show that the precision of QTL mapping increases as the number of F_2:3 families or the number of plants of each F_2:3 family increases. When the product of the number of F_2:3 families and the number of plants of each F_2:3 family is fixed, two sampling strategies could be studied. The results from simulation experiments show that the number of F_2:3 families can provide more information than the number of plants of each F_2:3 family.Once the observations of F₂ plants and F_2:3 family average for a quantitative trait were measured, joint analysis of F₂ and F_2:3 populations is better than a single-population analysis.3) Bayesian mapping of epistatic QTL in the F_2:3 design. Provided that epistasis between any two markers is included in the genetic model, Bayesian analysis can be used to estimate all QTL effects. The results from simulation experiments show the same trend described above.