Host-associated microbiomes are shaped by networks of interactions among microbes and between microbes and the host. It is not possible to directly measure these interactions, requiring inference from co-occurrence data. Such inferences are prone to high rates of false positives and negatives, and the complications found in real data - in particular demographic noise, higher-order and frequency-dependent interactions among microbes, and variation across hosts - are not well understood. In this project a direct test will be developed using the utility of co-occurrence-based interaction networks in understanding host-associated microbial communities where these biologically realistic complications are present. Further, the PIs will test the common implicit assumption that variation in individual hosts within a population does not fundamentally alter the inter-species interactions in a microbiome, such that the same interaction network can be used to understand the microbiome. Educational and outreach activities are specifically designed to excite and engage students at all levels in biophysics and theoretical biology. The proposed work will actively engage undergraduate students from the Biology and Physics programs in independent biophysics research. Further, quantitative laboratory modules are being developed for the Biology and Physics curricula at Emory to familiarize students with core concepts in population dynamics and to introduce the concepts of variation and hidden states in biological data.
The PIs will use the nematode worm Caenorhabditis elegans and an eight-species bacterial consortium from the worm's native microbiome. The worm is a powerful model system for understanding microbial community assembly, where bottom-up assembly and quantification microbiome composition in large numbers of homozygous individual hosts is possible on short experimental time scales, allowing rapid generation of large, high-quality data sets describing microbiome composition and variation. Using this tractable system, the PIs will obtain microbiome composition data from large numbers of individual hosts, directly parameterizing interactions among microbes (competition, facilitation etc.) and between individual microbial species and the host (e.g. colonization rates and densities in the intestine). Using these data, they will fit and constrain neutral and near-neutral stochastic models of community assembly to: 1) determine the interactions shaping these systems, 2) measure the effects of higher-order and frequency-dependent interactions on inferred networks, and 3) test the assumption that microbial interaction networks are conserved across hosts. The project will use a tractable experimental system as a tool to understand a complex network inference problem on noisy data with hidden underlying states, a problem that is highly relevant for real- world understanding of microbiome data and other data from systems where interactions are not directly measurable, and is therefore relevant for the Physics of Living Systems program. The combined expertise of the PIs will allow rigorous analysis of the technical problem while maintaining a firm connection of these analyses and conclusions to the biological reality.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.