High-quality and universal empirical atomic charges for chemoinformatics applications

Area of Science:

• Computational Chemistry
• Chemoinformatics
• Drug Discovery

Background:

• Partial atomic charges are crucial for understanding molecular behavior and are widely used in chemoinformatics.
• Existing Electronegativity Equalization Method (EEM) parameter sets have limited applicability due to missing parameters for essential elements and bond types found in drug-like molecules.

Purpose of the Study:

• To develop new EEM parameter sets that overcome the limitations of existing ones.
• To enable accurate and efficient calculation of partial atomic charges for a broader range of drug-like molecules.

Main Methods:

• Developed six novel EEM parameter sets.
• Validated charge calculation quality against established quantum mechanics (QM) methods (HF/6-311G, B3LYP/6-311G) and charge schemes (MPA, NPA, AIM).
• Assessed applicability across major drug databases (DrugBank, ChEMBL, Pubchem, ZINC).

Main Results:

• The new EEM parameters yield charge calculation quality comparable to QM methods.
• Parameters demonstrate excellent performance on both training and test datasets.
• Achieved applicability for at least 95% of molecules in key drug databases, a significant improvement over existing parameters (<60%).

Conclusions:

• Developed EEM parameters facilitate rapid, high-quality partial atomic charge calculations for nearly all drug-like molecules.
• A user-friendly software solution for computing EEM parameters is provided, enabling direct application in chemoinformatics.
• This work expands the utility of EEM for advanced chemoinformatics tasks.