MIMEco: Multi-objective metabolic modeling to predict and explain pairwise interactions

Abstract

Genome-scale metabolic models (GEMs) provide a mechanistic framework for quantitatively exploring an organisms physiology and its interactions with the environment. However, their use in microbial community modeling has been limited since it usually relies on community-wide objective functions or the integration of community abundance data. Nevertheless, predicting microbial interactions is crucial for understanding ecosystem dynamics, and thus requires dedicated methods. Here, we introduce MIMEco (Metabolic Interaction Modeling in Ecosystems), an open-source Python package that predicts pairwise microbial interactions through multi-objective optimization. MIMEco does not require abundance data, provides a user-friendly workflow, and predicts interaction types, strengths, and exchanged metabolites directly from Pareto fronts. In this study, MIMEco is validated by replicating experimental results of an in vitro co-culture of two auxotrophic Escherichia coli strains. By combining mechanistic inference with usability, MIMEco makes metabolic interaction modeling accessible to non-specialists while remaining flexible for experts, establishing a versatile platform for advancing metabolic ecology across biomedical and environmental fields. IMPORTANCEDeciphering microbial interactions through in vitro experiments is often costly and limited in scope, while existing bioinformatics tools can be complex to use or limited by actual sample data. MIMEco offers an open-source and user-friendly multi-objective framework for predicting and quantifying pairwise microbial ecosystem interactions and nutrient exchanges. We demonstrate its reliability by replicating a well-characterized laboratory experiment, showing that MIMEco can accurately reflects real biological behavior. This tool improves the communitys ability to explore and understand metabolic interaction mechanisms underpinning microbial ecosystems.