Metabolic Interactions and Capabilities within Microbial Communities 394 views

The collective action of microbes that colonize humans, known as their microbiota, has emerged as a major force influencing health and disease. Appreciation for this influence has grown exponentially in recent years thanks to advances in high-throughput sequencing technologies that were coupled with key concepts in microbial ecology and evolution linking sequence to phylogeny. Although studies utilizing these technologies, and the analytic methods also required to enable them, have greatly advanced our understanding of the potential impact of the microbiota on human health, we still lack systematic methods for interrogating the mechanisms responsible for the numerous associations we have identified. In this dissertation, I present my work which accelerates our ability to develop actionable hypotheses from associations observed in studies of the microbiota. This work is divided into data-driven approaches for inferring metabolic mechanisms governing interspecies interactions (Chapter 2), and model-driven approaches for improving our understanding of metabolism for gut microbes (Chapters 3 and 4). In Chapter 2, I developed a method that establishes expected metabolic behavior within microbial communities based on the assumption of constant metabolite yield between mono- and co-culture conditions. Using this method, I identified global improvements in the efficiency of biomass production that occur in co-cultures in which a species experienced a growth benefit, and show that the method can be used to interrogate complex interspecies interactions such as cross feeding. In Chapters 3 and 4, I developed software and method for building and analyzing ensembles of genome-scale metabolic network reconstructions. I used these tools to address a key challenge in systems biology: how should curation of large mechanistic models be prioritized? I developed an ensemble generation, simulation, and analysis framework that identifies key curation targets that maximally reduce simulation uncertainty, thereby establishing the relative value of curating each portion of a metabolic network. In sum, the work in this dissertation represents advances in our ability to interrogate interactions between gut microbes as well as our ability to efficiently construct predictive models of metabolism for individual species.

PHD (Doctor of Philosophy)

Computational Biology; Microbiome; Metabolism; Metabolic Modeling

English

2019-04-16