(536g) [Invited Keynote] Towards More Accurate Models of Microbiome Metabolism: Integrating Theory and Data

Metabolic modeling has been increasingly shown to be a useful approach in predicting and understanding microbiome metabolism. Usually, metabolic networks are reconstructed from metagenomes or reference genomes for the detected microorganisms. Metabolite consumption and production of individual members are quantitatively simulated using methods derived from flux balance analysis (FBA) that implements the mass balance principle and constrains reaction directionality. This provides a picture of possible interactions within the microbiome and roles of individual microbes detailed to the level of metabolic reactions and metabolites.

Despite the promise, there are critical challenges in generating accurate predictions from FBA-based models. One challenge is that FBA was intended for single organisms and omits important ecological and evolutionary principles for multi-organism communities. We will discuss our progress in integrating the concept of Nash equilibrium into the FBA framework for improving predictions from the theory side. It predicts important interactions such as cooperation vs non-cooperation that are not captured by FBA and shows significantly improved growth rate predictions for a large-scale yeast-algae coculture experiment. Another challenge is that FBA is usually used as a parameter-free model, which is a strength but at the same time means that there is not a standard systematic procedure for parameterizing a model to improve predictions given observed data. In another study, we developed a scheme for parameterizing a model for an anaerobic digestion community to improve model predictions from the data side. The results suggest that the fermentation profile significantly constrains the range of allowable community membership and metabolism. Based on the model, potential strategies to perturb the community for maximizing volatile acid production can be identified.