MetabolEx integrates expert-based transformation rules with machine learning classifiers trained on 41,000+ experimentally validated reactions spanning 194 metabolic transformations. Covering Phase I and Phase II metabolism, the platform delivers probabilistically ranked predictions with automated toxicophore screening and interactive metabolic network visualization.
Machine learning classifiers rank metabolites by reaction probability, reducing false positives compared to exhaustive rule-based enumeration.
Generate metabolite trees up to three recursive levels with interactive network visualization for immediate pathway exploration.
Probabilistic filtering achieves 6-fold higher precision than rule-only methods at deep prediction levels, maintaining chemical plausibility under combinatorial complexity.
Explore metabolic pathways through interactive node-edge visualizations with transformation metadata, reaction probabilities, toxicophore alerts, and export-ready publication diagrams.
Curated set of 41,000+ experimental reactions across 194 transformations spanning 8 metabolic classes, exhaustively mapping critical Phase I and Phase II pathways including bioactivation routes.
Automated toxicophore screening via ProfhEX integration flags bioactivation liabilities and reactive metabolites for early-stage drug safety assessment.