Conditional copula modeling for the analysis of multiple phenotypes in genetic association studies

Genome-wide association studies (GWAS) aim to identify genetic markers associated with various human traits. However, many genetic markers are linked to multiple phenotypes, reflecting the complexity of genetic influence and the concept of pleiotropy. Traditional single-phenotype approaches may overlook broader genetic effects across correlated traits. This dissertation addresses these challenges by extending the method of Konigorski et al. (2019) to develop a conditional copula-based framework for estimating and testing the dependence between phenotypes conditioned on a genetic effect. Our approach models the dependence structure between traits relative to a genetic marker, improving the precision of genetic effect estimation. Through simulation studies, we demonstrate that our model behaves well under estimation and testing procedures, providing unbiased estimates, controlling Type I error rates, and yielding good statistical power. This work contributes to advancing GWAS methodologies by addressing correlations between traits, enhancing the understanding of complex genetic architectures, and offering an alternative tool for multi-phenotype genetic analysis.