New prediction methods for solubility parameters based on molecular sigma profiles using pharmaceutical materials.

Area of Science:

• Pharmaceutical Science
• Computational Chemistry

Background:

• Solubility parameters are crucial in chemical and pharmaceutical sciences.
• Accurate prediction of solubility parameters from chemical structure is highly desirable.
• In silico methods offer a promising avenue for predicting these parameters.

Purpose of the Study:

• To compare the predictive accuracy of two in silico methods (COSMO-RS and QSPR) for solubility parameters of pharmaceutical compounds.
• To validate these in silico methods against an experimental dataset, including new measurements.
• To assess the performance of classical group contribution methods for comparison.

Main Methods:

• Utilized the conductor-like screening model for real solvents (COSMO-RS) and quantitative structure-property relationship (QSPR) models.
• Employed screening charge densities of molecular surfaces (σ-profiles) for in silico predictions.
• Experimentally determined solubility parameters using inverse gas chromatography for validation.

Main Results:

• Initial evaluation on reference solvents showed excellent predictions: R² of 0.855 (COSMO-RS) and 0.945 (QSPR).
• For pharmaceutical compounds, prediction accuracy was good: R² of 0.701 (COSMO-RS) and 0.717 (QSPR).
• Classical group contribution methods demonstrated lower predictive power compared to the in silico approaches.

Conclusions:

• The in silico methods, COSMO-RS and QSPR, are highly promising for predicting pharmaceutical solubility parameters.
• Challenges in predicting solid-state pharmaceuticals were noted, impacting accuracy.
• These computational tools can significantly aid in drug discovery and formulation processes.