Prediction of octanol-water partition coefficients using a group contribution solvation model

Using our recently derived group contribution solvation (GCS) model, we hav e developed a predictive model for the octanol-water partition coefficient (K-OW), the GCSKOW model. In this model K-OW is calculated from two molecul ar structure parameters, which take into account the size and shape effects , and one energy parameter that determines the attractive interactions betw een the solute and the solvent. On the basis of quantum mechanical studies, we found that for organic solutes with a single strong functional group, a ll these parameters can be obtained in a group contribution manner. Consequ ently, we present a database here for various functional group contribution s in this new, easy-to-use model. The root-mean-square deviation of the pre dicted log K-OW from the GCSKOW model for 226 solutes is found to be 0.14 ( which corresponds to 38% in K-OW), which is considerably less than those fr om the methods of Hansch and Leo (0.18 in log K-OW or 51%), KOW-UNIFAC (0.2 1 or 62%), and LSER (0.23 or 71%).