(279g) Quantitative Structure-Property Relationship Model for Prediction of Octanol-Water Partition Coefficient

The octanol-water partition coefficient (K_ow) is a thermophysical property describing the hydrophobicity of a molecule as expressed by the ratio of the organic and aqueous solubilities. This property has been found useful in inferring transport mechanisms and molecular distribution in environmental areas and bio-applications. For example, K_ow is a significant property in describing the permeation of a compound through the skin in transdermal drug delivery.

In this study, we present a new quantitative structure property relationship (QSPR) model for predicting K_ow utilizing a wide range of molecular species data from the PHYSPROP physical property database. Using the random diversity sampling technique, which insures a diverse sampling of the molecular species in the database, 5598 and 4144 molecules were assigned to training and validation sets, respectively. Non-linear QSPR models involving robust neural network algorithms with non-linear descriptor reduction techniques based on differential evolution were developed.

We improve on other literature models in several aspects, including (a) use of multiple QSPR software to provide descriptors assuring both model superiority and stability, (b) automated optimization of molecular structure conformations, (c) a wholly nonlinear descriptor reduction strategy, and (d) use of robust non-linear neural networks with multiple initializations to ensure network stability. Employing these state-of-the-art nonlinear algorithms, we have modeled accurately log K_ow with a root mean square error of 0.24 and 0.39 in the training and validation sets, respectively. Further, our models compare favorably to those available in literature using a standardized, literature test set of molecules.