Prediction of pKa values using the PM6 semiempirical method

Area of Science:

• Computational Chemistry
• Physical Organic Chemistry

Background:

• Predicting acid-base properties (pKa) is crucial in chemistry and drug discovery.
• Accurate pKa prediction aids in understanding molecular behavior and optimizing drug candidates.

Purpose of the Study:

• To evaluate the accuracy of the PM6 and PM6-D3H+ methods with continuum solvation models for pKa prediction.
• To compare these semiempirical methods against ab initio results and empirical predictors.

Main Methods:

• Utilized PM6 and PM6-D3H+ methods with SMD and COSMO solvation models.
• Employed isodesmic reactions for pKa calculations of pyridines, alcohols, phenols, benzoic acids, carboxylic acids, and amines.
• Compared results with published ab initio data and an empirical pKa predictor.

Main Results:

• PM6 methods achieved accuracy comparable to ab initio calculations for most compound classes (0.5-1.0 pH units MAD).
• PM6 predictions correctly identified the protonation state of a drug-like molecule at physiological pH, unlike an empirical predictor with a large error.
• The computational cost is low, enabling high-throughput screening applications.

Conclusions:

• PM6-based methods offer a computationally efficient and accurate approach for pKa prediction.
• These methods are suitable for high-throughput screening in drug discovery and chemical research.
• PM6 provides a reliable alternative to more computationally expensive ab initio methods for pKa determination.