Breaking the Barriers: Machine-Learning-Based c-RASAR Approach for Accurate Blood-Brain Barrier Permeability...

Area of Science:

• Computational chemistry and cheminformatics
• Drug discovery and development
• Pharmacology and toxicology

Background:

• Predicting blood-brain barrier (BBB) permeability is critical for central nervous system (CNS) drug efficacy and safety.
• Existing models face challenges in accurately assessing drug permeability across the BBB.
• The development of novel computational approaches is needed to improve CNS drug design.

Purpose of the Study:

• To introduce and validate a classification read-across structure-activity relationship (c-RASAR) framework for enhanced BBB permeability prediction.
• To showcase the refinement of BBB permeability prediction for organic compounds using the c-RASAR approach.
• To provide a versatile computational platform for assessing neuropharmacological implications in drug development.

Main Methods:

• Developed a machine learning-based c-RASAR linear discriminant analysis (LDA) model using 7807 compounds from the B3DB database.
• Integrated principles from read-across and quantitative structure-activity relationship (QSAR) methodologies.
• Validated the model using three external datasets: natural products (LOTUS), drug-like compounds (DrugBank), and FDA-approved drugs.

Main Results:

• The c-RASAR framework demonstrated a powerful capability in predicting BBB permeability.
• The model accurately assessed the permeability of diverse compound sets, including natural products and approved drugs.
• Machine learning-based c-RASAR models showed significant improvements in predictive accuracy.

Conclusions:

• The c-RASAR framework offers a refined and accurate method for predicting BBB permeability.
• This approach advances the understanding of molecular factors influencing CNS drug permeability.
• The computational platform facilitates rapid compound assessment, aiding informed decision-making in drug development.