Insights into the permeability of drugs and drug-likemolecules from MI-QSAR and HQSAR studies

Abstract

Membrane-interaction QSAR (MI-QSAR) and Holographic QSAR (HQSAR) analyses have been performed on a diverse set of drugs and drug-like molecules. MI-QSAR combines a set of membrane-solute interaction properties calculated during molecular dynamics simulation with the set of classical solute descriptors to predict the biological behavior of drugs and drug-like molecules. HQSAR is a technique which employs fragment fingerprints or molecular holograms as predictive variables of biological activity. A data set of 60 structurally diverse molecules with permeability coefficients were used to construct significant MI-QSAR and HQSAR models of Caco-2 cell permeation. A statistically meaningful MI-QSAR model was obtained with r (2) = 0.805 and q (2) = 0.696. Subsequently, HQSAR models were developed on the same data set. The best HQSAR model (r (2) = 0.915, q (2) = 0.539) was obtained with fragment distinctions atom, bond, donor and acceptor with atom count 4 to 7. The predictions for training and test set molecules are in good agreement with experimental results and show the potential of models for untested compounds. This displays the importance of MI-QSAR and HQSAR analysis in estimating ADME properties characterized by the transport of solutes through biological membranes.