In Silico Prediction of Membrane Permeability from Calculated Molecular Parameters.

A data set consisting of 712 compounds was used for classification into two classes with respect to membrane permeation in a cell-based assay:  (0) apparent permeability (P_app) below 4 × 10^(-6) cm/s and (1) (P_app) on 4 × 10^(-6) cm/s or higher. Nine molecular descriptors were calculated for each compound and Nearest-Neighbor classification was applied using five neighbors as optimized by full cross-validation. A model based on five descriptors, number of flex bonds, number of hydrogen bond acceptors and donors, and molecular and polar surface area, was selected by variable selection. In an external test set of 112 compounds, 104 compounds were classified and 8 compounds were judged as “unknown”. Among the 104 compounds, 16 were misclassified corresponding to a misclassification rate of 15% and no compounds were falsely predicted in the nonpermeable class.