SophosQM: Accurate Binding Affinity Prediction in Compound Optimization.

Area of Science:

• Computational Chemistry
• Drug Discovery
• Molecular Modeling

Background:

• Accurate prediction of compound-target binding affinity is vital for efficient drug development.
• Current methods often face challenges in balancing accuracy, reliability, and speed.
• Predicting binding energies computationally can significantly accelerate the drug discovery pipeline.

Purpose of the Study:

• To introduce SophosQM, a novel quantum mechanics (QM)-based approach for predicting compound binding affinities.
• To demonstrate the accuracy and reliability of SophosQM in estimating binding energies for drug-like molecules.
• To provide a computationally efficient method for high-throughput screening in drug discovery.

Main Methods:

• Utilized the fragment molecular orbital (FMO) method to compute the enthalpic contribution to binding free energy.
• Incorporated clogP as a macroscopic descriptor to approximate the entropic component of binding.
• Employed multilinear regression to correlate FMO-derived enthalpic terms and clogP with experimental binding affinity data.
• Validated the method's performance on 70 compounds across six pharmaceutically relevant protein targets.

Main Results:

• Achieved a high global correlation coefficient of approximately 0.9 between predicted and experimental binding affinities.
• Demonstrated satisfactory predictive performance comparable to free energy perturbation methods.
• Showcased the potential for high-throughput application using semiempirical QM methods, reducing computation time to seconds per compound.

Conclusions:

• SophosQM offers an accurate, reliable, and fast QM-based method for predicting compound-protein binding affinities.
• The approach effectively integrates quantum mechanical calculations with macroscopic descriptors for robust predictions.
• SophosQM presents a valuable tool for accelerating drug discovery through efficient virtual screening and lead optimization.