DOI

https://doi.org/10.25772/DVH0-PZ52

Author ORCID Identifier

https://orcid.org/0000-0002-9396-2860

Defense Date

2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy

Department

Human and Molecular Genetics

First Advisor

Michael C. Neale, PhD

Second Advisor

Roseann E. Peterson, PhD

Abstract

Genetic epidemiology advances our understanding of how genetic factors influence health and disease in the population, including psychiatric and behavioral traits, and helps elucidate the causal processes underlying the observed epidemiological associations. This dissertation focuses on two methodological areas in this rapidly evolving field: cross-population genomic discovery and causal inference, with applications to two particularly burdensome public health problems: depression and smoking.

Genome-wide association studies (GWASs) are extremely valuable in identifying the genomic loci associated with complex traits and have identified hundreds of genomic risk loci for depression. However, as is the case for most human genetic studies, extant GWASs of depression predominantly comprise individuals of European descent, limiting the generalizability to non-European populations and exacerbating health disparities. Moreover, the large-scale genetic studies have primarily examined shallow, minimally defined depression phenotypes, with low specificity to clinical major depressive disorder (MDD). Aim 1 of this dissertation addressed these limitations through improvements in trans-ancestry analytical approaches and data with strictly defined MDD outcomes, identifying novel genomic risk loci for MDD and highlighting the genetic differences between MDD and broad depression. This work provides a publicly shareable analytical pipeline to facilitate inclusive, cross-population genetic studies, helping improve genomic discovery, global representation, and generalizability.

Beyond the genetic etiology of a single trait, understanding the causal processes between co-occurring traits is fundamental to disentangling the causes and consequences of these traits. In observational studies, causality may be inferred using family/twin data, longitudinal data, or the associated genetic loci. However, all three approaches rely on different assumptions that, if violated, can lead to biased results. Aim 2 leveraged recently developed models combining genetic and twin data to avoid some of these assumptions and uncovered unidirectional and bidirectional causal influences between cigarette smoking and epigenome-wide DNA methylation. This study illuminates the role of DNA methylation in both the susceptibility to and the adverse effects of smoking. Longitudinal causal models may fail to detect causation if the time interval is too long between the observed cause and effect. Therefore, Aim 3 combined genetic and longitudinal data to estimate additional shorter-term effects that do not persist for the length of the interval. An application of this new model identified shorter-term effects of alcohol consumption on smoking, along with more persistent reverse effects, which may partly contribute to the concurrent smoking and alcohol consumption. Finally, Aim 4 developed novel models integrating twin and longitudinal data to differentiate between longitudinal confounding and causation, as well as between short-term and lagged effects. Applications of these models indicated bidirectional causal effects between smoking and depression in young adults, suggesting a reinforcing feedback loop in the emotion-smoking comorbidity.

In summary, this dissertation introduces several methodological innovations for genomic discovery and causal inference, contributing to our understanding of the genetic etiology of MDD and the causal links between smoking and its biochemical (DNA methylation) and behavioral (depression and alcohol use) correlates.

Rights

© The Author

Is Part Of

VCU University Archives

Is Part Of

VCU Theses and Dissertations

Date of Submission

8-4-2025

Available for download on Saturday, August 03, 2030

Share

COinS