When Trees Guide Molecules: Multiobjective Search in de Novo Drug Design.

Area of Science:

• Computational chemistry and cheminformatics
• Drug discovery and development
• Machine learning in medicinal chemistry

Background:

• Molecular design faces challenges in exploring vast chemical spaces for bioactive compounds.
• Discovering molecules with selective or polypharmacological inhibitory profiles is a key hurdle in drug development.

Purpose of the Study:

• To present a computational framework for de novo generation of drug candidates.
• To optimize molecules targeting orexin and histamine receptors with desired selectivity or multitarget activity.
• To integrate guided exploration strategies for efficient chemical space navigation.

Main Methods:

• Integration of Monte Carlo tree search with predictive machine learning models.
• Utilization of expert-defined rules for de novo molecule generation.
• Multiobjective optimization including inhibitory action, toxicity, aqueous solubility, and melting temperature.

Main Results:

• Demonstration of a hybrid approach for efficient exploration of chemical space.
• Generation of inhibitor candidates with optimized pharmacological profiles and properties.
• Scalable pathway for in silico drug design toward selective or multitarget agents.

Conclusions:

• The developed framework enables efficient navigation of chemical space for drug discovery.
• The hybrid approach facilitates the design of molecules with desired selectivity or polypharmacological profiles.
• This method offers a scalable solution for in silico optimization in drug design.