Antimicrobial resistance (AMR) is a leading global health threat and is currently among the top causes of death worldwide. If left unaddressed, AMR is projected to cause 10 million deaths annually by 2050. AMR arises when microorganisms such as bacteria, fungi and viruses develop the ability to resist the effects of antimicrobial drugs intended to eliminate them. This resistance can emerge naturally through genetic mutations that confer survival advantages, and it can also spread through horizontal gene transfer, allowing bacteria to share resistance genes within microbial communities. The misuse and overuse of antimicrobials exacerbate this problem by increasing selective pressure, accelerating the acquisition and spread of antimicrobial resistance genes (ARGs). The decreasing cost and increasing accuracy of whole-genome sequencing have made it a vital tool in AMR surveillance. It is now recommended by the World Health Organization for monitoring resistance globally. Whole-genome sequencing enables comprehensive analysis of entire microbial communities, allowing for the rapid and efficient detection of ARGs across populations. Traditional surveillance efforts often focus on clinically important pathogens with high resistance potential, such as the ESKAPEE group and target AMR hotspots including hospitals and wastewater treatment plants. While these cross-sectional efforts help identify resistance genes and assess prevalence, they provide only a single snapshot in time. The absence of temporal data limits their predictive value and restricts the ability to identify emerging resistance patterns before they escalate into outbreaks. Additionally, these approaches overlook healthy populations, limiting the understanding of the healthy resistome and how it changes over time. To address these limitations, this study introduces a new pipeline that uses longitudinal metagenomic data to categorize antimicrobial resistance genes based on their temporal behavior within microbial environments. This dynamic approach integrates existing analytical tools to annotate ARGs, mobile genetic elements (MGEs), predicted toxins and taxonomic information in order to assess the dissemination potential of ARGs and investigate how short-term and long-term perturbations affect their dynamics. The pipeline was first applied to a longitudinal dataset of healthy individuals to examine ARG dynamics in the gut microbiome. Four ARG categories were identified, dynamic, persistent, transient and stagnant, based on inflow and outflow probabilities calculated by the pipeline. These categories were then analyzed for their association with MGEs, predicted toxins and taxonomic annotations. To extend this analysis, the pipeline was applied to two additional longitudinal cohorts that reflect different types of microbiome disruption. Short-term disruption was represented by fecal metagenomic data from individuals colonized by carbapenemase-producing Enterobacteriaceae (CPE) and their family members. Since CPE colonization often resolves without intervention, this cohort allowed for comparison of ARG behavior under transient colonization. Long-term disruption was evaluated using a cohort of individuals with liver cirrhosis who were either untreated or administered rifaximin. This cohort enabled the investigation of both chronic disease and antibiotic effects on the gut microbiome and also included saliva samples that allowed assessment of ARG dynamics in the oral microbiome. Overall, this pipeline introduces a temporal surveillance approach for antimicrobial resistance genes. By moving beyond traditional static analyses, it provides a comprehensive framework to examine the behavior of ARGs over time, their relationships with MGEs, toxins and taxonomic groups, and their potential to emerge as future threats to public health.
Flux-Based Modeling of Antimicrobial Resistance Gene Mobility in the Human Gut and Oral Microbiome
Hill, E. (Author). 1 Dec 2025
Student thesis: Doctoral Thesis › Doctor of Philosophy