QSAR study on water solubility of drugs and VolSurf Descriptor

There exists an important relationship between the water solubility of drugs and their absorption. In this study, VolSurf, a novel computerized method, is used for predicting water solubility of drugs and determining the critical molecular properties needed for water solubility. The tested molecules consist of 26 diverse compounds. Using partial least squares discriminant analysis to correlate the experimental data with the theoretical molecular properties of water solubility, a good model (r(2) = 0.90, q(2) = 0.77) was obtained. The correlative analysis after the compounds are divided into training and test sets shows that the model from the former of 18 compounds has a good predictivity on the other 8 compounds and the standard deviation error of prediction (SDEP) is 0.59. Descriptor analysis demonstrates that the lower best three local energy minima of interaction between molecule and water probe and larger distance between them are favorable to water solubility. The molecule in which hydrophilic effect dominates has high solubility, while the compound with strong hydrophobic property is converse. At the same time, the solubility of large molecule is lower than that of the small one.