Improving Succinic Acid Production in Actinobacillus Succinogenes through Genome-Scale Metabolic Models
Conference on Constraint-Based Reconstruction and Analysis (COBRA)
2018
5th Conference on Constraint-Based Reconstruction and Analysis (COBRA 2018)
Poster Session
Poster Session
Sunday, October 14, 2018 - 6:00pm to 7:00pm
Identifying specific microbial drivers of variation in metabolic phenotypes is a major goal in the study of host-associated and environmental microbiomes. Correlation-based analysis of paired microbiome-metabolome datasets is a widespread approach to this objective. To date, however, the efficacy and limitations of this approach have not been evaluated. To address this challenge, we developed a mathematical definition of the contribution of each taxon to metabolite variation based on its uptake and secretion fluxes. We applied a multi-species dynamic genome-scale metabolic modeling pipeline to simulate simplified gut communities, generating idealized microbiome-metabolome datasets with precisely known microbial metabolic fluxes. Comparing the observed taxon-metabolite correlations in this simulated setting with the true taxonomic contributors, we found that correlation-based analysis poorly predicts key contributors, with low accuracy and a high false discovery rate. Importantly, however, the predictive value was strongly influenced by both metabolite and taxon properties, as well as exogenous environmental variation. These findings have practical implications for the analysis and interpretation of microbiome-metabolome studies. This study illustrates the utility of multi-scale metabolic modeling as a simulation tool to inform study design and methods development in microbiome research.