Ligand-Conditioned Side Chain Packing for Flexible Molecular Docking.

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Drug Discovery

Background:

• Molecular docking is essential for understanding protein-ligand interactions.
• Machine learning (ML) docking methods show promise but struggle with pose plausibility and protein flexibility.
• Existing methods have limitations in real-world applications.

Purpose of the Study:

• To introduce ApoDock, a novel modular docking paradigm.
• To address the challenges of physical plausibility and protein flexibility in ML-based docking.
• To improve the accuracy and reliability of molecular docking predictions.

Main Methods:

• ApoDock combines ML-driven conditional side chain packing with traditional sampling.
• It ensures physically realistic poses by considering protein backbone and ligand data.
• A mixture density network-based scoring function is used for pose ranking.

Main Results:

• ApoDock demonstrates competitive performance across various applications.
• It shows a 28.5% higher success rate than state-of-the-art methods when using modeled structures (e.g., AlphaFold2, ESMFold).
• The method achieves accurate side chain packing, physical-based pose sampling, and reliable pose ranking.

Conclusions:

• ApoDock offers a robust solution for generating physically plausible protein-ligand docking poses.
• Its ability to handle protein flexibility and utilize modeled structures makes it highly valuable.
• ApoDock represents a significant advancement for protein-ligand binding studies and drug development.