Pre-docking filter for protein and ligand 3D structures

Area of Science:

• Computational chemistry
• Drug discovery
• Structural biology

Background:

• Virtual drug screening (VDS) is computationally intensive, requiring significant resources for binding affinity calculations.
• Millions of chemical compounds are available for VDS, necessitating efficient methods to prioritize candidates.
• Identifying and removing low-affinity ligands early can accelerate the VDS procedure.

Purpose of the Study:

• To develop and evaluate a pre-docking filtering strategy to enhance the efficiency of virtual drug screening.
• To investigate the relationship between ligand properties (volume, rotatable bonds) and binding affinity.
• To reduce the computational burden of VDS by filtering out unlikely binders.

Main Methods:

• Performed protein-ligand docking for 6353 ligands against 21 protein X-ray crystal structures.
• Ranked docked ligands by calculated binding affinities.
• Analyzed the characteristics (volume, rotatable bonds) of top-ranked and bottom-ranked ligand sets.

Main Results:

• Top-ranked ligands exhibited volumes and rotatable bond counts similar to those in crystal structures and matched binding site volumes.
• Bottom-ranked ligands were often too large to fit or too small for specific, high-affinity binding.
• Ligand volume and flexibility correlated with binding site characteristics and affinity.

Conclusions:

• A pre-docking filter considering binding site and ligand shape/volume can effectively remove low-affinity compounds.
• This filtering approach significantly increases the speed of virtual drug screening.
• Optimizing ligand selection prior to docking improves the efficiency of drug discovery pipelines.