Molecular Environment-Specific Atomic Charges Improve Binding Affinity Predictions of SAMPL5 Host-Guest Systems

Area of Science:

• Computational Chemistry
• Molecular Modeling
• Biophysics

Background:

• Host-guest systems are crucial for benchmarking computational methods in predicting binding free energies.
• Accurate force fields are essential for reliable binding affinity predictions in alchemical free energy calculations.
• Current methods for deriving atomic charges can limit the accuracy of these predictions.

Purpose of the Study:

• To develop a novel methodology for deriving atomic charges in host-guest systems.
• To improve the accuracy of absolute binding free energy predictions.
• To assess the impact of quantum mechanics/molecular mechanics-derived charges on host-guest binding affinity.

Main Methods:

• Utilized quantum mechanics/molecular mechanics (QM/MM) calculations with minimal basis iterative stockholder (MBIS) partitioning.
• Developed an interface between OpenMM and ORCA software for D-MBIS charge derivation.
• Calculated the energetic cost of guest polarization in both bound and unbound states.

Main Results:

• D-MBIS charges, when used with the general Amber force field, enhanced binding affinity predictions for six guests in octa acid hosts.
• The new method outperformed the AM1-BCC charge set, particularly after correcting for polarization energy differences.
• The polarization cost correction significantly impacted the binding affinity predictions for anionic guests.

Conclusions:

• The D-MBIS charge derivation methodology offers improved accuracy for host-guest binding free energy calculations.
• Accounting for the energetic cost of guest polarization is critical for accurate predictions, especially for charged species.
• This work provides a more robust computational approach for studying host-guest interactions and molecular recognition.