Explaining blood-brain barrier permeability by synergistic effect on molecular substructures

Abstract

The blood-brain barrier (BBB) permeability is a critical factor in designing drug candidates intended either to act within the central nervous system (CNS) or to restrict distribution to the CNS. However, explaining BBB permeability requires accounting for multiple underlying mechanisms and diverse molecular substructures that influence this property. This complicates developing a fully interpretable understanding of BBB permeability. In this study, an explainable machine learning algorithm is proposed to BBB permeability based on synergistic effects on molecular substructures. Furthermore, our approach can be applied to mechanisms related to BBB permeability. Synergistic groups that positively or negatively influence the target property can be identified through a relative importance analysis of each substructure. This allowed us to screen for molecules with multiple positive or negative effects on the target property. Our approach can provide both explainable and predictive models for the design of drug candidates. This record is sourced from MEDLINE/PubMed, a database of the U.S. National Library of Medicine