A surface-integral model for log P OW.

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Drug discovery

Background:

• The octanol-water partition coefficient (Log P(OW)) is crucial for drug design.
• Existing models for Log P(OW) prediction have limitations.

Purpose of the Study:

• To develop a novel surface-integral model for calculating Log P(OW).
• To improve the accuracy and applicability of Log P(OW) prediction in drug design.

Main Methods:

• Utilized AM1 semiempirical molecular orbital theory to calculate local surface properties.
• Developed a new hydrophobicity calculation scheme based on binning local surface properties.
• Trained and validated the model on extensive datasets of chemical compounds.

Main Results:

• The developed model demonstrates performance comparable to or exceeding leading commercial Log P(OW) prediction models.
• A variant of the model effectively handles flexible compounds using Boltzmann-weighted conformational predictions.
• This represents the first 3D QSPR model leveraging conformational ensembles on a large dataset.

Conclusions:

• The surface-integral model offers a powerful and accurate method for Log P(OW) prediction.
• The model's ability to incorporate conformational flexibility enhances its utility in drug design.
• This approach advances the field of quantitative structure-property relationship (QSPR) modeling.