The Covalent-MAS framework has been released for iterative drug design

Developers introduced Covalent-MAS — a modular Python framework designed to automate the process of creating covalent drugs. The tool is focused on solving problems of iterative molecular design using machine learning methods.

The system supports multi-criteria optimization and evidence-based training. The framework includes an adapter for AutoDock Vina, which allows for molecular docking to evaluate the affinity of compounds. The use of this stack made it possible to increase the valid molecule generation rate to 98.81%.

The solution architecture is built on a modular principle, giving researchers the flexibility to customize pipelines for generating and filtering chemical structures. The framework is designed for integration into existing computational environments used in pharmaceutical development and bioinformatics.

The project is published in open access for specialists in computational chemistry. The tool allows reducing the time spent on hypothesis testing when searching for new active compounds due to automated iterative data processing.