A Ranged Series of Drug Molecule Fragments Defining Their Neuroavailability

Abstract

Computer modelling methods based on “structure-property” relationships (QSPR) allows rational planning of both experimental studies and the effective generation of structures with the ability to cross the blood-brain barrier (BBB). The aim of this work was to create a “structure-property” model based on the simplex presentation of molecules for non-experimental assessment of their ability to cross the BBB and to carry out structural interpretation of the model, taking account of the mutual influences of atoms in the molecule. The simplex method for representing molecular structures was used to evaluate the contributions of different molecular fragments and functional groups to the level of neuroavailability of compounds diffusing across the BBB. The presence of strongly polar and ionogenic groups (carboxyl, carbonyl, phenol hydroxyl) was found to have adverse effects on the ability of substances to penetrate into the brain. The presence of halogen atoms and aromatic fragments had positive influences on transfer of substances through the BBB. The QSPR model constructed here was characterized by satisfactory predictive accuracy as assessed using the RandomForest internal model validation procedure (R2 = 0.54; RMSE = 0.47).