MetabolEx: platform for small molecule metabolites generation and metabolic pathway prediction
Year: 2025
Type: article
Abstract: Accurate prediction of metabolite structures is crucial for understanding drug disposition, anticipating safety liabilities, and guiding medicinal chemistry to optimize the pharmacokinetic properties of lead compounds. In this work, we developed MetabolEx, a new computational platform to predict metabolites formed during drug metabolism, encompassing both phase I and phase II reactions. MetabolEx combines a rule-based engine, which enumerates all possible metabolites, with a machine learning model which estimates the probability associated with each predicted metabolite. While the rule-based approach enables an exhaustive exploration of all possible metabolic pathways, the probability score allowed to reasonably restrict the number of generated metabolites maintaining biochemical plausibility. MetabolEx was trained on a comprehensive dataset, including data from both public and proprietary databases enabling it to capture both common and rare metabolic pathways. The results revealed a good capacity for MetabolEx to correctly predict metabolites formed by both phase I and phase II reactions. In addition, it reaches the state-of-the-art tool BioTransformer in terms of performance, with the advantage of reducing the false positive rate as more reaction levels are explored. MetabolEx is freely available as a webserver at https://metabolex.exscalate.eu.
Authors: Lunghini, Filippo, Agazzi, Giada, Tazzari, Valerio, Fava, Anna, Vittorio, Serena, Vistoli, Giulio, Beccari, Andrea Rosario
Institutions: Dompé (Italy), University of Milan
Language: English
Subfield: Computational Theory and Mathematics
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7105666869
FWCI:
Cites:
Cited by:
Related to: