A Priori Prediction of Tissue: Plasma Partition Coefficients (Log BP) of Drugs to Facilitate the Use of MLR and MLR-GA Methods

It is important to determine whether a candidate molecule is capable of penetrating the plasma-brain barrier indrug discovery and development. The aim of this paper is to establish a predictive model for plasma-brainbarrier penetration using simple descriptors The usefulness of the quantum chemical descriptors, calculated atthe level of the DFT and HE theories using 6-310* basis set for QSAR study of anti-vial NucleosideAnalogues drugs was examined. Delivery of anti-viral agents into the central nervous system (CNS) is clmicallyhoportant. Nucleoside analogues are a major source of clinically used antiviral agents. The QSAR modeldeveloped contributed to a mechanistic understanding of the investigated biological effects. The first step in thisstudy was to use a dauset containing 23 drugs with known activity. In the next steps some of them with thelarge secondary chain branches were removed to make our approach. Multiple Linear Regressions (MLR) wasemployed to model the relationships between molecular descriptors and biological activities of molecules usingstepwise method and genetic algorithm as variable selection tools. Biological activities contain the logarithm ofthe ratio of the steady-state concentration of a compound in the brain to in the plasma, log Bp. A multiparametricequation containing maximum six descriptors at HF/6-3 10* and eight descriptors at 133LYP/6-3 10*method with good statistical qualities(Rpax— 0.976 , 11214n— 0.959 at HF/6-3/G* and Rwax4 0.979 , 112KGP 0.952 at B3LYP/6-316*) wasbtiledbyMl(plLR gr g teP thoi Th Sal d d ibis p pe ppears to bevery simple but robust and effective for predictive use This method relates log Bp values to fundamentalmolecular properties, such as Electrostatic Potential. Local charge, Electric Field Gradient, Isotropicparameters, Natural Population Analysis. Also, GA-MLR regression was used to model the structure — activityrelationships.