Integrating Quantum Mechanics into Protein-Ligand Docking: Toward Higher Accuracy and Reliability.

Area of Science:

• Computational Chemistry
• Structural Biology
• Drug Discovery

Background:

• Traditional flexible ligand docking methods often misrepresent intramolecular relative energies of conformations.
• This inaccuracy leads to significant errors in conformational energies and unrealistic ligand geometries with high strain energies.

Purpose of the Study:

• To introduce QFVina and QFVinardo, novel protein-ligand docking methods.
• To enhance the accuracy of docking-based virtual screening by incorporating precomputed, high-quality conformational libraries.
• To address the inaccuracies in intramolecular relative energies and strain energies inherent in flexible ligand docking.

Main Methods:

• Leveraging precomputed conformational libraries with quantum mechanics (QM)-optimized geometries.
• Utilizing *ab initio* DFT-D4-based conformational rankings and strain energy calculations.
• Implementing physically realistic ligand conformations with accurate strain energies into scoring functions.

Main Results:

• Demonstrated substantial errors in conformational relative energies for widely used force field-based methods.
• Showcased that traditional flexible ligand docking produces geometries with significant strain energies, comparable to binding energies.
• QFVina and QFVinardo produced markedly different docking results compared to traditional methods, even with identical interaction energy scoring functions.

Conclusions:

• Accurate representation of ligand conformations and strain energies is critical for reliable protein-ligand docking.
• The developed QFVina and QFVinardo methods offer improved accuracy in virtual screening.
• Findings highlight the limitations of current flexible ligand docking approaches and the necessity of physically realistic conformational data.