Further development and validation of empirical scoring functions for structure-based binding affinity prediction.

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Estimating protein-ligand binding affinity is crucial for drug discovery.
• Accurate scoring functions are needed for predicting binding free energies.
• Existing methods often require refinement for improved accuracy.

Purpose of the Study:

• To develop novel empirical scoring functions for protein-ligand binding affinity.
• To implement and evaluate multiple algorithms for hydrophobic effect calculations.
• To create a consensus scoring function (X-CSCORE) for enhanced prediction accuracy.

Main Methods:

• Developed empirical scoring functions incorporating van der Waals, hydrogen bonding, deformation, and hydrophobic terms.
• Implemented three distinct algorithms for the hydrophobic effect calculation.
• Calibrated scoring functions using multivariate regression on 200 protein-ligand complexes.
• Combined three scoring functions into a consensus function, X-CSCORE.
• Validated X-CSCORE on an independent set of 30 protein-ligand complexes.

Main Results:

• Scoring functions reproduced training set binding free energies with standard deviations of 2.2, 2.1, and 2.0 kcal/mol.
• X-CSCORE predicted binding free energies on an independent set with a standard deviation of 2.2 kcal/mol.
• X-CSCORE demonstrated improved molecular docking accuracy compared to conventional force field methods.

Conclusions:

• The developed empirical scoring functions, particularly the consensus function X-CSCORE, provide accurate predictions of protein-ligand binding affinity.
• X-CSCORE offers significant improvements in molecular docking accuracy.
• X-CSCORE holds potential for advancing drug discovery and development processes.