Metabolic Network Modeling and Metabolomics Integration for Comparative Analysis of Biogeochemical Reactions in Multiple River Systems

Abstract:

River corridors form biogeochemical hotspots that control the material and energy cycles in the natural environment. Data-model integration is critical for reliable prediction of temporal and spatial variations of biogeochemical reactions in these spots. Collection of omics profiles from globally distributed river corridors through the Worldwide Hydrobiogeochemical Observation Network for Dynamic River Systems (WHONDRS) consortium provides valuable resources in this regard. Full integration of omics data into conventional lumped models is however ineffective due to their oversimplified representation of reaction pathways. In this work, we demonstrate the utility of genome-scale metabolic models in improving our understanding of biogeochemical reactions. Through a week-long summer school hosted by EMSL at PNNL in July 2020, we developed genome-scale metabolic networks across seven different samples covering four river systems based on WHONDRS data, using the DOE’s KBase (www.kbase.us) modeling pipeline. We constructed metabolic networks of metagenomes and their high-quality extracted bins. This allowed for comparative analysis of community- and individual taxon-level metabolism and biogeochemical potentials, e.g., in regard to differential representation of anaerobic pathways in pore and surface water samples. Significant differences in metabolic functions across river systems were predicted by incorporating high-resolution metabolomics obtained from Fourier transform ion cyclotron resonance mass spectrometry. We used KBase chemoinformatics tools to expand the known chemistry of reference compounds through iterative addition of reactions while pruning to avoid a combinatorial explosion of potential compounds and reactions. The resulting models generate a comprehensive description of potential biogeochemical reactions with unprecedented detail.