Multivariate Methods for the Study of Beta-lactam Resistance in Streptococci

Antibiotic resistant bacteria are a major source of mortality worldwide and are set to become one of most pressing threats to public health in the 21st Century. Consequently, over the past two decades, there have been several efforts to conduct large-scale surveillance of resistance in circulating bacterial populations. While these datasets have provided a rich source of information to study drug resistance, they have also highlighted its complex, multivariate nature, involving correlations both within and between drug classes. These studies have driven the development of methods for the analysis of genotypes, but tools for the study of resistance phenotypes have remained limited, particularly with regard to quantitative, multivariate traits. To bridge this gap, this thesis develops multivariate methods into a framework that can model high-dimensional drug resistance in large collections of isolates. By applying these tools to beta-lactam resistance in streptococci, I demonstrate these methods can improve visualisation and modelling of complex phenotypes, assist in identifying the molecular basis of multivariate change, and aid in studying repeatable patterns in the evolution of these traits. Rather than focusing on single traits, this novel approach emphasises the multidimensional pattern across phenotypes, leading to new insights into beta-lactam resistance in streptococci. More generally, the methods provide a conceptual framework to study multivariate drug resistance in pathogens, and tie together several facets of drug resistance evolution into a concise visual representation, which can be more easily interpreted by researchers and public health bodies.

BBSRC Doctoral Training Partnership