Parallel multi-swarm cooperative particle swarm optimization for protein-ligand docking and virtual screening

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• High-quality molecular docking is crucial for efficient drug development.
• Existing methods like AutoDock Vina and its variants show room for performance improvement.
• Novel docking programs aim for enhanced speed and accuracy in molecular modeling.

Purpose of the Study:

• To develop a novel parallel multi-swarm cooperative particle swarm model for protein-ligand docking.
• To improve the accuracy, robustness, and efficiency of molecular docking and virtual screening.
• To enhance the identification of drug-like active compounds.

Main Methods:

• Implementation of a parallel multi-swarm cooperative particle swarm model.
• Utilizing a master-slave swarm architecture with mutual cooperation and co-evolution.
• Employing a random drift particle swarm optimization algorithm within the multi-swarm framework.

Main Results:

• The multi-swarm model demonstrates superior docking robustness compared to PSOVina.
• Achieved top-tier accuracy in protein-ligand docking and significant enrichment of drug-like compounds.
• Exhibited the second-best AUC screening accuracy with lower computational demands than most competitors.

Conclusions:

• The proposed multi-swarm cooperative model offers a novel algorithmic approach for protein-ligand docking and virtual screening.
• Coevolution between master and slave swarms enables remarkable parallel docking performance.
• The source code is publicly available for further research and application.