Prediction of Protein-Ligand Binding Affinities Using Atomic Surface Site Interaction Points

Area of Science:

• Computational chemistry
• Structural biology
• Drug discovery

Background:

• Atom surface site Interaction Points (AIP) have successfully predicted binding constants in synthetic systems.
• Extending AIPs to protein-ligand interactions requires novel descriptor generation and matching techniques.

Purpose of the Study:

• To adapt and validate the Atom surface site Interaction Points (AIP) method for predicting protein-ligand binding free energies.
• To assess the accuracy of AIP-based free energy calculations against experimental data.

Main Methods:

• Protein binding site AIPs were generated using precomputed descriptors and graph-based substructure matching.
• Ligand AIPs were derived from molecular electrostatic potential surfaces calculated via density functional theory.
• Intermolecular AIP contacts were identified by projecting ligand and protein AIPs onto X-ray crystal structures.

Main Results:

• The method successfully identified intermolecular interactions based on spatial proximity of AIPs.
• The overall free energy of binding was computed by summing AIP contact and desolvation contributions.
• Application to 94 uncharged ligand complexes in the CASF dataset yielded a Pearson correlation coefficient of 0.76 and an RMSD of 11 kJ mol⁻¹.

Conclusions:

• The extended AIP method provides a robust approach for predicting protein-ligand binding free energies.
• This computational strategy holds promise for accelerating drug discovery and development.
• Accurate prediction of binding free energies is crucial for understanding molecular recognition events.