Prediction of Membrane Permeation of Drug Molecules by Combining an Implicit Membrane Model with Machine Learning

• DOI: 10.1021/acs.jcim.8b00648
• Publication Date: 2018-12-12T17:56:58Z
• Authors: Stephanie A. Brocke; Alexandra Degen; Alexander D. MacK...; Bercem Dutagaci; Michael Feig
• Keywords: Partial charge; Membrane permeability; Synthetic membrane

Lipid membrane permeation of drug molecules was investigated with Heterogeneous Dielectric Generalized Born (HDGB)-based models using solubility-diffusion theory and machine learning. Free energy profiles were obtained for neutral by the standard HDGB Dynamic (DHDGB) to account deformation upon insertion drugs. We also hybrid free where neutralization charged taken into insertion. The evaluation predictions done against experimental permeability coefficients from Parallel Artificial Membrane Permeability Assays (PAMPA), effects partial charge sets, CGenFF, AM1-BCC, OPLS, on performance discussed. (D)HDGB-based improved over two-state implicit models, sets seemed have a strong impact predictions. Machine learning increased accuracy predictions, although it could not outperform physics-based approach in terms correlations.