Investigating molecular features that influence blood - brain barrier permeability of compounds

Introduction: The blood-brain barrier (BBB) is a semipermeable border that is responsible for maintaining central nervous system (CNS) homeostasis in the brain. Screening compounds based on their BBB permeability is an important consideration for CNS-acting drug development. Several studies have attempted to link physicochemical properties to specific BBB permeability; however, there is no widely accepted rule that can determine how and to what extent molecular properties affect the BBB permeability of compounds. To understand the complex phenomenon of BBB permeability, we explored the vast molecular space of the compounds to determine whether some features could differentiate the compounds based on their BBB permeability.Materials and Methods: A dataset of 1951 compounds containing 1246 BBB-permeable and 705 BBB-nonpermeable molecules was used in the study. Each compound encoded 499 molecular features. Feature selection was performed using feature selection algorithms, feature-to-feature, and feature-to-permeability correlation analysis. The findings of the feature selection algorithms were merged to select the best set of 53 features.Results: The descriptive analysis of the molecular features suggests that nCXr (number of X on ring C[sp3]) feature values for BBB nonpermeable compounds are zero for all considered compounds except for compounds with PubChem ID 71260, Flurithromycin. In addition, the majority of compounds were found to have nCrq (number of ring quaternary C[sp3]) feature values of zero for BBB nonpermeable compounds. For BBB-permeable compounds, MACCS fingerprints 8 feature values for all 1951 compounds were found to be zero except for the compound with PubChem ID 146291, Dezinamide.Conclusion: The descriptive and nonparametric tests confirm that molecular feature distributions are significantly different for BBB permeable and BBB nonpermeable compounds.