GWAS and the depths of variation: From locus discovery to risk prediction

This thesis explores methodological, biological, and translational aspects of genome-wide association studies (GWAS). It begins with a historical overview of statistical genetics, followed by a primer on the GWAS methodology. Subsequently, it presents post-GWAS approaches for biological interpretation, focusing on gene-mapping, functional annotation, and convergent pathway analyses. Empirical chapters investigate Alzheimer’s disease genetics through the largest multi-ancestry GWAS to date, address methodological biases in GWASs of polygenic score-derived phenotypes, and examine local genetic sex differences across quantitative traits. Finally, the thesis introduces a Bayesian framework to transform polygenic scores into directly interpretable disorder probabilities, thereby enhancing their clinical utility. Together, these contributions advance our understanding of Alzheimer’s disease genetics, genetic sex differences, refine methods for robust inference, and highlight translational opportunities for polygenic score prediction.