XenoNet

XenoNet predicts metabolic pathways and intermediate metabolites to assess formation of reactive metabolites in drug metabolism.

Key Features:

• Metabolic Network Prediction: Enumerates metabolic networks including intermediate metabolites to capture bioactivation pathways relevant to reactive metabolite formation.
• Pathway Enumeration: Accepts a substrate (initial drug molecule) and a target product (final metabolite) and enumerates sequences of intermediate structures between them.
• Likelihood Computation: Computes likelihood scores for each predicted pathway and intermediate metabolite to prioritize probable routes.
• Validation and Performance: Validated on 17,054 metabolic networks derived from experimental literature, achieving pathway recall of 88%, intermediate metabolite recall of 46%, top-one pathway accuracy of 93.6%, and intermediate metabolite accuracy of 51.9%.

Scientific Applications:

• Risk Mitigation: Identifying potential reactive intermediates that could lead to adverse drug reactions.
• Drug Design Optimization: Informing structural modifications to reduce formation of harmful metabolites.
• Streamlining Drug Development: Prioritizing compounds and pathways computationally to reduce reliance on experimental assays during early-stage drug development.

Methodology:

Takes substrate and target product inputs, enumerates intermediate metabolite pathways, computes likelihood scores for pathways and intermediates, and validates predictions against experimental metabolic networks (17,054 networks).