Predicting aqueous solubility by QSPR modeling.

Area of Science:

• Computational chemistry
• Cheminformatics

Background:

• Accurate prediction of aqueous solubility is crucial in drug discovery and environmental science.
• Quantitative Structure-Property Relationship (QSPR) models offer a computational approach to estimate physicochemical properties.

Purpose of the Study:

• To develop robust QSPR models for predicting aqueous solubility.
• To investigate the utility of combining Gibbs energy of solvation and sublimation descriptors with the octanol-water partition coefficient for solubility prediction.

Main Methods:

• Utilized quantitative structure-property relationship (QSPR) modeling.
• Employed descriptors for Gibbs energy of solvation and sublimation.
• Incorporated octanol-water partition coefficient.
• Applied artificial neural networks for model refinement.

Main Results:

• All developed QSPR models achieved R-squared > 0.7 and Q-squared > 0.6 for aqueous solubility prediction across seven datasets.
• Uncoupling descriptors for Gibbs energy of sublimation improved model performance.
• Artificial neural network refinement further enhanced model accuracy and reduced standard deviation.

Conclusions:

• The combination of specific descriptors and partition coefficients provides a reliable strategy for QSPR-based aqueous solubility prediction.
• Artificial neural networks offer a powerful tool for refining QSPR models, leading to improved predictive performance.