Alkane/Water Partition Coefficient Calculation Based on the Modified AM1 Method and Internal Hydrogen Bonding...

Area of Science:

• Computational Chemistry
• Physical Chemistry
• Molecular Modeling

Background:

• Accurate prediction of partition coefficients is crucial for drug discovery and chemical process design.
• Existing models often struggle to account for complex molecular interactions like internal hydrogen bonds (IHBs).

Purpose of the Study:

• To develop and validate a novel physics-based model for calculating solute partition coefficients between water and alkanes.
• To incorporate the effects of internal hydrogen bonds (IHBs) on partition coefficients.
• To provide a computationally efficient method for predicting molecular properties.

Main Methods:

• A semi-empirical method combining COSMO (COnductor-like Screening Model) charge density calculation with a bond-correction algorithm.
• Statistical sampling to account for the conformational states of internal hydrogen bonds (IHBs).
• Validation against experimental partition coefficient data for approximately 3500 molecules, including small organics and drug-like compounds.

Main Results:

• The model achieved a root-mean-square deviation of approximately one log 10 unit when compared to experimental partition coefficients.
• Analysis revealed that IHBs can exist in distinct 'open' (in water) and 'closed' (in apolar solvents) states.
• Failure to account for IHB state changes can introduce significant prediction errors (up to two log 10 units per IHB).

Conclusions:

• The developed physics-based model offers a significant improvement in predicting partition coefficients by explicitly addressing IHBs.
• The computational cost is manageable (minutes per molecule on a single core), making it suitable for screening applications.
• This approach bridges the gap between faster, less accurate methods (QSAR) and slower, more computationally intensive simulations.