Abstract
Computer-aided methods are applied to the development of an estimation method for aqueous solubilities of organic compounds. Multivariate linear regression models in terms of various molecular descriptors, in particular molecular size and polar descriptors, are examined. The proposed model has been tested by predicting the aqueous solubilities of about 500 organic compounds having diverse structures and functionalities. For the majority of compounds studied, good agreement between observed and estimated solubility is confirmed. This study gives new insight into the effect of atomic charges on the solubility of organic compounds in water.