Calculation of Aqueous Solubility of Organic Compounds from Molecular Structural Properties

Abstract

The quantitative structure-activity relationships (QSAR) correlations were developed for the aqueous solubility of organic compounds at 298.15, 473.15, and 573.15 K. To elucidate the molecular structural variables to control the solvation phenomena, we firstly developed the QSAR correlation of solubility of 108 organic molecules in water at 298.15 K with the use of experimentally and theoretically derivable variables. The experimental variable used was the octanol-water partition coefficient and melting point, while the theoretical variables were LUMO energy and total molecular surface area, which were determined with the COSMO-PM3 method. Based on the results at 298.15 K, the QSAR correlations for aqueous solubility at 473.15 K and 573.15 K were determined with the use of the literature solubility data for 21 compounds at these two high temperatures. It was found that the QSAR correlation with COSMO-PM3 method is able to predict the aqueous solubility of organic compounds at 473.15 and 573.15 K from the molecular structural properties and the melting temperature, since the octanol-water partition coefficient can be predicted from atom/fragment contribution method.

Application of Random Forest and Multiple Linear Regression Techniques to QSPR Prediction of an Aqueous Solubility for Military Compounds

“…In analyzing the results in the QSPR studies listed in Table 1, several important disadvantages of a traditional approach can be seen. In studies16, 19, 22, 23 only the small data sets of specific classes of organic molecules were still used. All the models listed in Table 1 are well‐fitted and most of them are robust.…”

“…In studies [16,19,22,23] only the small data sets of specific classes of organic molecules were still used. All the models listed in Table 1 are well-fitted and most of them are robust.…”

“…However, these models were mainly devoted to drug or drug-like compounds and thus cannot be used for solubility prediction of military compounds, which are structurally dissimilar. Finally, for computational methods that are based on a direct quantum-mechanical evaluation of solubility, [19][20][21][22][23] it is still necessary to use physicochemical characteristics as descriptors and to employ potentially time-consuming software for solubility modeling. Sometimes such approaches cannot process large and diverse sets of compounds.…”

Calculation of Solubility of Organic Compounds in Supercritical CO2 Based on Quantitative Structure-Activity Relationships-Development of Correlation Equation Using Experimental Variable-

“…The COSMO-PM3 method gave relatively good results compared to the PM3 method. We achieved good correlations for aqueous solubility of 108 organic compounds − 27 − using the quantitative structure-activity relationship (QSAR) correlations with experimentally and theoretically derivable variables [4].…”