A Comparative Study of Transferable Aspherical Pseudoatom Databank and Classical Force Fields for Predicting...

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Accurate electrostatic interaction evaluation is critical for macromolecular chemistry and drug design.
• Quantum mechanical methods offer high accuracy but are computationally expensive.
• Efficient approximation methods are needed for large molecular systems.

Purpose of the Study:

• To evaluate the University of Buffalo Pseudoatom Databank (UBDB) approach for approximating electrostatic properties of macromolecules.
• To compare UBDB's performance against established quantum mechanical methods and popular force field models.
• To assess the accuracy and speed of UBDB for electrostatic energy calculations.

Main Methods:

• Utilized S66 and JSCH-2005 datasets (208 molecular complexes).
• Calculated reference electrostatic energies using Symmetry-Adapted Perturbation Theory (SAPT) at the B3LYP/aug-cc-pVTZ level.
• Compared UBDB results with reference energies and those from AMBER (AM1-BCC, RESP) and CHARMM (CGenFF) force fields.

Main Results:

• UBDB demonstrated high consistency with reference electrostatic energies (R² ≈ 0.98).
• UBDB showed improved accuracy over popular force field models (RMSE ≤ 3.2 kcal mol⁻¹ vs. 3.7-7.6 kcal mol⁻¹).
• Investigated discrepancies in electrostatic potentials and dipole moments among methods.

Conclusions:

• The UBDB approach offers an accurate and computationally efficient method for estimating electrostatic interaction energies.
• UBDB's performance suggests potential for broader applications in macromolecular simulations and drug design.
• UBDB provides a valuable alternative to computationally intensive quantum mechanical calculations and less accurate force field models.