Molecular size based approach to estimate partition properties for organic solutes

Abstract

A model is presented that achieves a remarkable reduction in the number of parameters used to predict the log octanol-water partition coefficient (log P) by using a three-dimensional estimate of molecular size. An algorithm combining analytical and numerical techniques is used to compute van der Waals molecular volume and surface area. Using this computed volume, a simple equation that adequately describes log P for a wide variety of organic compounds is obtained by introducing only one additional parameter. Its value is determined by the oxygen- or nitrogen-containing functional groups present in the molecule and correlates with hydrogen bond acceptor basicity. The corresponding free energy change agrees well with that accepted for hydrogen bonding in water. Tested on over 700 organic compounds comprising various pharmaceutically useful molecules, including even neutral peptides, this distinctively simple and fully computerized 3D approach compares favorably with empirical, complicated, "two-dimensional" fragment methods. The fact that here only a very limited number of parameters are used proves that these 2D fragment methods were acceptable only when computation of spatial molecular structures was computationally too demanding.