Improved Methods for Understanding the Gut Microbiome

The gut microbiome plays a profound role in human health. Over the last decade, much work has been done to define differences between microbial communities in healthy and diseased individuals. Studies of healthy individual have been small, and failed to address the multifactorial influences on the microbiome. However, little effort has focused on comparisons between the effects, in part due to the multivariate and complex nature of the data, which does not fit into the traditional paradigm for effect size calculations. This leaves questions about the relative impact of common practices on the microbiome, in comparison to the impact of a disease state. To address these challenges, I have combined novel and traditional approaches to microbiome analysis. I developed a method for estimating statistical power and effect size using Monte Carlo Simulation. The technique works as well as traditional approaches for parametric data, and out-performs the traditional methods when applied to nonparametric data. The observed effect sizes quantified previously observed biological conclusions. I applied the new power method to data from the American Gut Project, the largest open source, crowdfunded microbiome citizen-science project. The sample size within the American Gut made it possible to quantify and compare previously unobserved lifestyle effects on the microbiome. The effect of plant consumption on the microbiome was almost as large as the effect of antibiotic use in the last month, demonstrating the importance of diet in shaping microbial communities. The power technique was also applied to a study of Parkinson\u27s disease, to demonstrate the large effect associated with a Parkinson\u27s diagnosis. This work also represents the first time the influence of Parkinson\u27s disease status on the microbiome has been separated from the influence of Parkinson\u27s disease treatment. I propose a mechanism whereby the microbiome may be modulating dopamine production in Parkinson\u27s patients through a metabolite, butyrate. I then explore other examples of microbial metabolites modulating disease. Finally, I propose the use of effect size calculation to identify targets for mechanistic investigation, followed by the application of multi-omics techniques to examine underlying pathways