GutBug

GutBug predicts the biotransformation of biotic and xenobiotic molecules by human gut bacteria to identify bacterial metabolic enzymes, reaction centers, and outputs relevant to dietary, nutraceutical, and drug metabolism.

Key Features:

• Predictive capability: Predicts bacterial metabolic enzymes that biotransform xenobiotics and biotic molecules, providing complete Enzyme Commission (EC) numbers and identifying specific reaction centers.
• Modeling techniques: Applies machine learning algorithms, neural networks, and chemoinformatic techniques for prediction and analysis.
• Training data: Trained on 3,457 enzyme substrates and a curated database of 363,872 enzymes from approximately 700 gut bacterial strains.
• Accuracy and validation: Reports accuracies ranging from 0.78 to 0.97 across reaction classes and has been validated on 27 known microbiome-mediated molecules including complex polysaccharides, flavonoids, and oral drugs.

Scientific Applications:

• Dietary and pharmacological research: Enables investigation of microbiome-mediated metabolism to inform inter-individual and population-level variation in drug efficacy and dietary responses.
• Prebiotic and nutraceutical development: Identifies potential biotransformation pathways to support the design and evaluation of prebiotics and nutraceuticals.
• Drug design optimization: Provides predictions of gut bacterial biotransformation that can inform optimization of drug bioavailability and activity.

Methodology:

GutBug employs machine learning algorithms, neural networks, and chemoinformatic techniques trained on 3,457 enzyme substrates and a curated database of 363,872 enzymes from ~700 gut bacterial strains to predict enzyme EC numbers and reaction centers.