Toward Post-Hartree-Fock Accuracy for Protein-Ligand Affinities Using the Molecules-in-Molecules Fragmentation-Based...

Area of Science:

• Computational Chemistry
• Molecular Modeling
• Quantum Chemistry

Background:

• Large molecular systems like protein-ligand complexes present significant computational challenges for accurate electronic structure calculations.
• Existing post-Hartree-Fock methods struggle with the size and complexity of these systems, limiting their application.

Purpose of the Study:

• To benchmark the Molecules-in-Molecules (MIM) method for post-Hartree-Fock calculations on large molecular systems.
• To develop a strategy for layer/theory selection in MIM for protein-ligand complexes.
• To assess the computational efficiency and accuracy of the MIM method for binding energy calculations.

Main Methods:

• Utilized the three-layer Molecules-in-Molecules (MIM3) method combined with Density-Localized Pair-Natural-Orbital (DLPNO)-based post-Hartree-Fock methods.
• Implemented an energy term cancellation strategy within the supermolecular equation for enhanced computational efficiency.
• Employed diffuse or small Pople-style basis sets in the middle layer of the MIM approach.

Main Results:

• Achieved highly accurate protein-ligand binding energies with errors less than 1 kcal mol^-1.
• Demonstrated significant reductions in computational costs compared to traditional methods.
• Observed strong correlations between computed and experimental interaction energies (R² ≈ 0.90 for CDK2, R² ≈ 0.78 for BZT-ITK).

Conclusions:

• The MIM method, particularly MIM3 with DLPNO-based approaches, is effective and efficient for calculating binding energies in large molecular systems.
• Strategic selection of basis sets and layer/theory is crucial for accuracy and efficiency.
• This work provides practical insights and methodologies for computational studies of molecular interactions.