A Multi-Objective Molecular Generation Method Based on Pareto Algorithm and Monte Carlo Tree Search.

Area of Science:

• Computational chemistry
• Drug discovery
• Artificial intelligence in medicine

Background:

• Drug discovery is hindered by the challenge of optimizing multiple molecular objectives simultaneously.
• Current computational methods are limited in handling more than four objectives, restricting molecular design advancements.

Purpose of the Study:

• To introduce a novel method, Pareto Monte Carlo Tree Search Molecular Generation (PMMG), for multi-objective molecular design.
• To overcome the limitations of existing algorithms in high-dimensional objective spaces.

Main Methods:

• Utilized Monte Carlo Tree Search (MCTS) to efficiently explore chemical space and identify the Pareto Front.
• Employed simplified molecular input line entry system (SMILES) for molecular representation.
• Applied PMMG to optimize multiple drug discovery objectives, including binding affinity and drug-likeness.

Main Results:

• PMMG achieved a 51.65% success rate in simultaneously optimizing seven objectives, outperforming existing methods by 2.5 times.
• Demonstrated PMMG's capability in generating molecules with high docking scores for EGFR and HER2 targets.
• Generated molecules with favorable predicted drug-like properties.

Conclusions:

• PMMG effectively addresses the challenge of multi-objective optimization in molecular design.
• The method shows significant potential to accelerate drug discovery pipelines by handling numerous complex objectives.
• PMMG represents a substantial advancement in computational approaches for identifying novel drug candidates.