LAUSR

Carbon source-dependent control of bacterial growth is fundamental to bacterial physiology and survival. However, pinpointing the metabolic steps important for cell growth is challenging due to the complexity of cellular networks. Here, the elastic net model and multilayer perception model that integrated genome-wide gene-deletion data and simulated flux distributions were constructed to identify metabolic reactions beneficial or detrimental to Escherichia coli grown on 30 different carbon sources. Both models outperformed traditional in silico methods by identifying not just essential reactions but also nonessential ones that promote growth. They successfully predicted metabolic reactions beneficial to cell growth, with high convergence between the models. The models revealed that biosynthetic pathways generally promote growth across various carbon sources, whereas the impact of energy-generating pathways varies with the carbon source. Intriguing predictions were experimentally validated for findings beyond experimental training data and the impact of various carbon sources on the glyoxylate shunt, pyruvate dehydrogenase reaction, and redundant purine biosynthesis reactions. These highlight the practical significance and predictive power of the models for understanding and engineering microbial metabolism. Explainable machine learning models were developed to understand the effects of 30 different carbon sources on E. coli growth and their mechanisms down to the individual metabolic steps. A regression model of elastic net (EN) and a deep learning model of multilayer perception (MLP) were trained to learn patterns and relationships between the simulated flux distributions and gene-deletion dataset. The lack of interpretability of the traditional MLP model was overcome with the SHAP interpretation method. This approach outperformed the previous reaction deletion simulation by identifying non-essential growth-promoting reactions in addition to the essential reactions. The models’ practical application was demonstrated by experimentally validating the predicted reactions responsible for increased or decreased cell growth under different carbon conditions. A regression model of elastic net (EN) and a deep learning model of multilayer perception (MLP) were trained to learn patterns and relationships between the simulated flux distributions and gene-deletion dataset. The lack of interpretability of the traditional MLP model was overcome with the SHAP interpretation method. This approach outperformed the previous reaction deletion simulation by identifying non-essential growth-promoting reactions in addition to the essential reactions. The models’ practical application was demonstrated by experimentally validating the predicted reactions responsible for increased or decreased cell growth under different carbon conditions. Explainable machine learning models were developed to understand the effects of 30 different carbon sources on E. coli growth and their mechanisms down to the individual metabolic steps.