Bayesian Analysis of mtDNA Population Dynamics

Abstract

Mitochondria are energy-producing organelles in eukaryotic cells with their own genome (mtDNA), which exists in multiple copies per cell. This allows for the coexistence of wild-type and pathogenic variant mtDNA, known as heteroplasmy. When the proportion of pathogenic mtDNA exceeds a critical threshold, mitochondrial function is impaired. Clonal expansion refers to the increase in pathogenic mtDNA within a cell, potentially leading to dysfunction, but its underlying mechanisms remain poorly understood. Mathematical modelling has emerged as a powerful tool to investigate mtDNA dy namics, allowing simulation of long-term dynamics that cannot be measured experimen tally. However, models rely on assumptions, simplifying the biological system, and require parameter inference from limited and often tissue-specific data, complicating their inter pretation. This thesis employs advanced statistical methods to improve our understanding of mtDNA population dynamics. First, a Bayesian model estimates the proportion of blood cells that have reached wild-type homoplasmy, demonstrating that T cell differentiation into memory cells selectively reduces pathogenic mtDNA. Second, a Bayesian classifica tion model infers the proportion of skeletal muscle fibres with oxidative phosphorylation (OXPHOS) defects from OXPHOS protein abundance data, outperforming existing clas sification method. Finally, the practicalities of comparing theories of clonal expansion using OXPHOS deficient data and mathematical models is investigated using real and synthetic datasets. Overall, this work underscores the potential of mathematical models in studying clonal expansion while highlighting the challenges posed by limited and variable biological data. It presents novel techniques for inferring mtDNA dynamics and emphasizes the need for more comprehensive experimental data to refine model accuracy and interpretability.