Improved protein-ligand binding affinity prediction by using a curvature-dependent surface-area model.

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• The hydrophobic effect is crucial for protein-ligand interactions.
• Current scoring functions often simplify hydrophobic free energy calculations using surface tension, neglecting molecular shape.
• Molecular shape significantly influences binding affinity, as indicated by experimental and theoretical studies.

Purpose of the Study:

• To develop an improved empirical scoring function for protein-ligand binding.
• To incorporate molecular shape information into hydrophobic free energy calculations.
• To enhance the accuracy of predicting protein-ligand binding affinity.

Main Methods:

• Development of a novel curvature-dependent surface-area model.
• Implementation of this model into a new empirical scoring function named Cyscore.
• Validation using benchmark protein-ligand datasets (PDBbind).

Main Results:

• Cyscore demonstrates improved prediction accuracy for protein-ligand scoring.
• The curvature-dependent model effectively distinguishes different surface types (convex, planar, concave).
• Cyscore outperforms existing scoring functions in binding affinity correlation and ranking tests.

Conclusions:

• The curvature-dependent surface-area model offers a more accurate representation of hydrophobic interactions.
• Cyscore provides a valuable tool for enhancing protein-ligand interaction studies.
• This approach has the potential to advance drug discovery and design.