Date of Award

Fall 1-1-2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Public Health

First Advisor

Zhao, Hongyu

Abstract

Genome-wide association studies (GWAS) have identified thousands of genetic variants associated with complex traits and diseases, but most of these variants reside in non-coding regions, making their biological functions challenging to study and interpret. To investigate the potential roles of these variants, expression quantitative trait locus (eQTL) analyses associate genetic variants with gene expression, providing insights into their regulatory roles. Integrating eQTL results with GWAS summary statistics has motivated transcriptome-wide association studies (TWAS), which focus on testing associations between imputed genetically regulated gene expression (GReX) and complex traits, offering a powerful gene-level view of genetic contributions to disease. Despite substantial progress, several limitations remain. The overlap between GWAS signals and known eQTLs is limited, and eQTLs account for only a small fraction of trait heritability. One possible reason is that many eQTL effects are highly context- or cell-type-specific, which bulk tissue analyses may obscure. Moreover, single-cell RNA sequencing (scRNA-seq) technologies now enable the exploration of genetic regulation at single-cell resolution, but most current single-cell eQTL and TWAS methods analyze cell types separately, not able to leverage shared regulatory patterns. Additionally, the challenge of distinguishing causal from correlated variants within linkage disequilibrium (LD) blocks limits the resolution of both eQTL and TWAS analyses. This dissertation presents three major contributions to address these limitations and advance our understanding of genetic regulation and disease mechanisms. First, we evaluate the robustness of mediated expression score regression (MESC), a method designed to estimate the proportion of trait heritability mediated by gene expression. Through simulations and analyses of multiple traits, we reveal that violations of modeling assumptions and imperfect eQTL effect estimation might lead to biased mediated heritability estimates. In addition, we find that integrating functional annotations can improve these estimates. Second, we propose CASE, a novel fine-mapping framework for single-cell eQTLs. CASE jointly models shared and cell-type-specific genetic effects, overcomes challenges from LD and over-sharing error, and outperforms existing fine-mapping methods in both simulations and real datasets. It can also better captures cell type specificity and functionally enriched and disease-associated eQTLs. Finally, we propose UTMOST_fa, an extension of the UTMOST framework that integrates functional annotations to improve single-cell TWAS (scTWAS). UTMOST_fa adaptively prioritizes regulatory SNPs, improving imputation accuracy and the identification of disease-associated genes across cell types. Moreover, UTMOST_fa can identify more disease-associated genes in scTWAS analyses, revealing novel cell-type-specific genes for inflammatory bowel disease (IBD) and other diseases. In conclusion, our work provides methodological innovations and biological insights, enhancing our understanding of the molecular mechanisms linking genetic variation to complex traits.

Download

Share

COinS