The body is an ecosystem, containing a diverse and physiologically active microbiota (commensal bacteria, archaea, protists, fungi, and viruses), where the whole is greater than the sum of its individual parts. The gut microbiota is a powerful, and malleable, modulator of our phenotypes. Loss of key microbial diversity, changes in community structure, and undesirable evolutionary trajectories can result in disease states. 

However, it is challenging to integrate all of this complexity and make systems-scale predictions of functional outputs. Furthermore, even if we could rationally engineer gut ecosystem function, we do not yet understand when or how these functions alter the health state of the host. Our lab builds computational and experimental tools for exploring and manipulating host-microbe systems, mapping out how microbial, genomic, physiological, and behavioral variation converge to influence human health and disease. Our work sits at the confluence of microbial ecology, computational systems biology, and precision medicine.

Our first major research direction has been to define microbiome health through a host lens, by integrating data from dense, longitudinal, molecular phenotyping (genomes, metabolomes, proteomes, microbiomes, etc) from thousands of individuals. The aim of these multi-omic population studies has been to leverage a range of statistical modeling and machine learning approaches to identify the components of the host phenotype that are resonant with variation in the ecology of the gut. For example, we found that variation in the ecology of the gut was significantly associated with thousands of health and disease markers. Furthermore, we found that we could predict cross-sectional variation in the ecological diversity of the gut with 11 blood metabolites, revealing a surprisingly intimate connection between the small molecules in our blood and the microbes in our guts. Indeed, in a more recent study, we found that the gut microbiome is a stronger driver of the composition of the blood metabolome than the human genome. These exploratory analyses have expanded our understanding of how changes in our commensal microbiota may causally drive clinically relevant changes in the body.

Our second major research direction has been on leveraging existing knowledge to build systems-scale, mechanistic models that map ecosystem composition to functional outputs. Specifically, we have built a metagenome-scale metabolic modeling platform, called MICOM, that extends Flux Balance Analysis (FBA) to diverse, multi-species communities. MICOM employs a novel implementation of FBA, called cooperative-tradeoff FBA (ctFBA), that imposes ecologically reasonable constraints and results in quantitatively accurate predictions of bacterial growth rates and certain metabolic fluxes (that have been tested) in the human gut. We believe MICOM can be a valuable tool for designing personalised dietary, prebiotic, and probiotic interventions that optimize the metabolic output of the gut ecosystem. Ongoing work indicates that MICOM can be used to predict inter-individual variation in short-chain-fatty-acid (SCFA) production and to predict probiotic or pathogen colonization/engraftment.

Many of the complex, chronic diseases that plague our modern world cannot be easily treated with a single pill. We are all unique, with our own genomes, microbiomes, and life-histories, and we often respond to exposures in distinct ways. In the past few years, we have looked into some of these heterogeneous responses and found striking connections to the ecology of the gut. For example, we found that individuals aging healthily show a steady drift in the ecological composition of their guts, becoming more and more dissimilar to others in the population, and this drift was also associated with a rise in anti-inflammatory blood metabolites and a reduced risk of death over a four-year follow-up. In addition, we found that the gene-content of an individual’s gut microbiome can be used to predict weight loss responses during a healthy lifestyle intervention, identifying individuals who may need a more drastic intervention to successfully lose weight. As a final example, we recently found that the composition of the gut could be leveraged to predict patient responses to statins (cholesterol-lowering drugs), both in terms of on-target cholesterol-lowering and in terms of off-target effects on metabolism, like increased insulin resistance. This work provides exciting hints at novel therapeutic approaches for healthy aging, weight loss, and prescription drug personalisation, all mediated by the gut microbiome. Over the next decade, we plan to dedicate significant effort towards translating our discoveries into the real world, through clinical partnerships, trials, and commercialisation. Ultimately, we hope to build ecologically grounded therapeutics that make precision healthcare more effective and affordable for all of humanity.