Examination Date

Spring 5-8-2026

Degree

Thesis

Degree Program

Microbiology, Immunology and Parasitology

Examination Committee

Christopher Taylor, Guoshun Wang, Alison Quayle, Liz Simon, Hui-Yi Lin, David Welsh

Abstract

The human gut and vaginal microbiomes serve diverse functions required for maintaining and protecting mucosal sites. Alterations in vaginal microbiome communities contribute to the acquisition and growth of sexually transmitted infections (STIs), including bacterial vaginosis (BV), chlamydia, and human immunodeficiency virus (HIV), while alterations in the gut microbiome have been shown to contribute to the development of numerous diseases. Similarly, exposure to pathogens and toxins such as HIV and alcohol have been shown to alter microbiome composition, contributing to intestinal inflammation, intestinal barrier disruption, and detrimental health consequences such as liver disease and cardiovascular disease. Characterizing the relationship between human microbiomes, STIs, and health has been hindered by common limitations associated with microbiome studies, including intra- and inter-individual variability, sparsity, and limited functional characterization of gut and vaginal taxa. Multi-omic profiling and machine learning provide a powerful framework to address these limitations. Multi-omics approaches applied to microbiome data have linked microbiome remodeling to functional alterations, while machine learning approaches have been shown to handle noise, sparsity, and complex non-linear relationships effectively. We applied a diverse set of multi-omics and machine learning methods to further characterize vaginal microbiome remodeling leading up to incident bacterial vaginosis onset, the role of the vaginal environment in chlamydia pathogenesis, as well as HIV- and alcohol-associated gut microbial remodeling. The findings from this dissertation serve as groundwork aimed at advancing the treatment and detection of vaginal STIs as well as the development of therapies aimed at mitigating alcohol- and HIV-associated comorbidities. Additionally, the work outlined in this dissertation serves as a framework for the application of machine learning and multi-omics methods aimed at enhancing current microbiome analysis.

Download

Share

COinS