End-Point Affinity Estimation of Galectin Ligands by Classical and Semiempirical Quantum Mechanical Potentials

Area of Science:

• Computational Chemistry
• Structural Biology
• Drug Discovery

Background:

• Protein-ligand affinity prediction is crucial for drug discovery.
• Quantum mechanical potentials offer a more accurate approach but require validation.
• Galectin-3 is a relevant target in various diseases.

Purpose of the Study:

• To systematically evaluate modern force fields and semiempirical quantum mechanical (SQM) methods for protein-ligand affinity prediction.
• To validate these computational methods using a large dataset of galectin inhibitors against galectin-3.
• To analyze the impact of solvation models and entropic terms on prediction accuracy.

Main Methods:

• Collated over a thousand galectin inhibitors with known affinities.
• Docked inhibitors into galectin-3 and evaluated multiple force fields and SQM methods.
• Employed implicit solvation models and analyzed energy terms and entropic contributions.

Main Results:

• The best methods achieved a Pearson correlation of 0.7-0.8 between computed and experimental affinities.
• Performance varied among methods, particularly for structurally similar ligands and those forming halogen bonds.
• Solvation models significantly influenced prediction success; entropic terms showed mixed results.

Conclusions:

• Modern computational methods show potential for protein-ligand affinity prediction but require careful selection and validation.
• Halogen bonding presents a significant challenge for current computational approaches.
• Further refinement of computational workflows, including solvation and entropic effects, is necessary for improved accuracy.