A pose prediction approach based on ligand 3D shape similarity

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Molecular docking is crucial for predicting ligand-protein interactions.
• Pose prediction failures often stem from inadequate sampling or scoring inaccuracies.
• Accurate pose prediction is vital for rational drug design and understanding molecular mechanisms.

Purpose of the Study:

• To develop an improved molecular docking method for accurate ligand pose prediction.
• To address the challenge of insufficient sampling in traditional docking approaches.
• To enhance the reliability of pose prediction by incorporating 3D shape similarity.

Main Methods:

• Developed a novel pose prediction method utilizing 3D ligand shape similarity.
• Integrated ligand conformation placement based on highest shape similarity to known crystal structures.
• Refined poses using side-chain repacking and Monte Carlo energy minimization.
• Validated the method on CSARdock 2012 and 2014 benchmark datasets.

Main Results:

• Ligand 3D shape similarity effectively replaces extensive conformational and orientational sampling when a co-crystal structure is available.
• The method achieved top-ranking poses within 2 Å RMSD for 85.7% of test cases.
• A median RMSD of 0.81 Å was obtained, outperforming methods with extensive sampling.
• The approach demonstrated superiority over existing shape similarity-based methods.

Conclusions:

• 3D shape similarity is a powerful strategy for accurate molecular docking pose prediction.
• The developed method offers a more efficient and accurate alternative to traditional sampling-intensive techniques.
• This approach holds significant potential for accelerating drug discovery and structural bioinformatics research.