Multiple Trait Multiple Interval Mapping of Quantitative Trait Loci from Inbred Line Crosses

Quantitative trait loci;mapping;statistical method;multiple trait;inbred lines

Silva, Luciano Da Costa E ; Zhao-Bang Zeng, Committee Chair,Howard D. Bondell, Committee Co-Chair,Trudy F.C. Mackay, Committee Member,Sujit K. Ghosh, Committee Member,Silva, Luciano Da Costa E ; Zhao-Bang Zeng ; Committee Chair ; Howard D. Bondell ; Committee Co-Chair ; Trudy F.C. Mackay ; Committee Member ; Sujit K. Ghosh ; Committee Member

University:North Carolina State University

关键词: Quantitative trait loci;    mapping;    statistical method;    multiple trait;    inbred lines;   

Others  :  https://repository.lib.ncsu.edu/bitstream/handle/1840.16/4217/etd.pdf?sequence=1&isAllowed=y

美国|英语

来源: null

【 摘 要 】

Tremendous progress has been made in recent years on developing statistical methods for mapping quantitative trait loci (QTL) from crosses of inbred lines. Most of the recent research is focused on strategies for mapping multiple QTL and associated model selection procedures and criterion. In Chapter 1, we review the progress of research on QTL mapping on one and multiple trait by maximum likelihood and Bayesian methods.Although in many instances multiple trait are measured in the same subject, single traits analyses have been the main stream for the purpose of QTL identiﬠcation. However, single trait analyses do not take advantage of correlation between traits. Multiple trait analysis allows an investigator to assess the pattern of action of QTL on multiple trait, such as, testing the hypothesis of existence of pleiotropic QTL versus the hypothesis of close linked QTL affecting multiple trait, and testing the hypothesis of QTL by environment interaction.In Chapter 2, we proposed a statistical model for mapping multiple QTL affecting multiple trait, the multiple trait multiple interval mapping (MTMIM) model. We also developed a score-based threshold for assessing signiﬠcance level of QTL effects on multiple trait. Our MTMIM model provides a comprehensive framework for QTL inference in multiple trait, in which the score-based threshold is built in as an essential and elegant tool for computing the signiﬠcance level of effects of putative QTL in the genome-wide scan, therefore, allowing us to build a set of models containing multiple QTL.In Chapter 3, we empirically showed that the score-based threshold maintains the false discovery rate within acceptable levels and the multiple trait analysis can bring insights into the analysis of data for the purpose of QTL identiﬠcation. The analysis of data from an experiment with Drosophila showed the potential of our MTMIM model in delivering complementary information regarding the genetic architecture ofcomplex traits, such as, estimating QTL effects on a set of traits simultaneously, testing for the presence of pleiotropic QTL, and estimating the genotypic covariance between traits. A generalized expectation maximization Newton-Raphson (GEM-NR) algorithm for maximizing the likelihood function and estimating parameters in the MTMIM model was compared to the expectation-conditional maximization (ECM) algorithm. Empirical comparison showed that GEM-NR speeded up the convergence of likelihood function considerably when compared to the ECM algorithm, while still delivering stable estimates of parameters.In Chapter 4, we proposed analytical formulae to predict the length of conﬠdence interval for position of QTL and to predict shape of the LRT around the position of QTL in highly saturate linkage maps and multiple trait analysis using large sample theory. Our results generalize the results of Visscher and Goddard (2004) and they can be used to predict the length of conﬠdence interval for position of QTL with a hypothesized effect on multiple trait, for any given coverage probability. Our analytical formulae can also be used to predict shape of LRT around the position of QTL. Furthermore, we proposed an alternative method for predicting the length of conﬠdence interval for position of QTL, the adjusted method. The adjusted method accounts for the length of the chromosome in which the QTL is located and can deliver more accurately predictions than the method with no adjustments, especially for QTL of low heritability. Our simulation results showed that for sample size of 300 and QTL with heritability levels of 5, 10 and 15%, there are good agreement between lengths of conﬠdence intervals empirically estimated and analytically predicted with the adjusted method.