Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Genetic applications using structural equation modeling
Abstract
This dissertation consists of three papers in two different parts. The first part is about multi-trait quantitative trait locus (QTL) mapping where we develop two new methods to account for the causal relationships among these multiple traits. The regression-based multi-trait structural equation model (SEM) method is presented in the first paper. This method is based on two steps, deriving the conditional expected genotypic value at a given position from the flanking marker genotypes, and then constructing the SEM by treating the coefficients of the genetic effect as observed values. This approach produces results very similar to maximum likelihood (ML) and simplifies computation. However, the regression-based method is inferior to mixture model ML, but ML is very difficult to evaluate under the SEM. To deal with this problem, we proposed a Bayesian approach to mixture SEM in the second paper. Parameters are estimated based on Markov Chain Monte Carlo (MCMC) sampling. The number of QTLs affecting traits is determined by the Bayes factor. The performance of both methods is evaluated by simulation and applied to data from a wheat experiment. Compared with single trait analyses, our proposed methods not only improved statistical power of QTL detection and precision of parameter estimates but also provided important insight into how genes regulate traits directly and indirectly by fitting a more biologically realistic model. In the second part (third paper), we proposed a two-level SEM to identify gene-environment interactions in the development of coronary heart disease (CHD) using the Framingham Heart Study data, The approach accounts for complex relationships among intermediate risk factors (hypertension, blood lipids, and blood glucose) and CHD (Level 1), and family structure (Level 2). Our approach has several advantages over classical methods: (1) it provides important insight into how genes and contributing factors affect CHD by investigating the direct, indirect, and total effects; and (2) it aids with development of biological models that more realistically reflect the underlying mechanism of development of CHD.
Subject Area
Statistics
Recommended Citation
Mi, Xiaojuan, "Genetic applications using structural equation modeling" (2009). ETD collection for University of Nebraska-Lincoln. AAI3350451.
https://digitalcommons.unl.edu/dissertations/AAI3350451