Finding Needles in a Haystack: Determining Key Molecular Descriptors Associated with the Blood-brain Barrier Entry of...

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Pharmacokinetics

Background:

• The blood-brain barrier (BBB) restricts the entry of many chemicals into the central nervous system.
• Accurate prediction of BBB penetration is vital for drug development and toxicology.
• Quantitative Structure-Activity Relationship (QSAR) models can predict chemical properties based on molecular structure.

Purpose of the Study:

• To develop and validate QSAR models for predicting the blood-brain barrier (BBB) entry of 415 chemicals.
• To evaluate the predictive performance of two distinct sets of molecular descriptors and their combination.
• To assess the impact of variable selection on QSAR model accuracy for BBB penetration.

Main Methods:

• Calculation of two sets of molecular descriptors (579 and 198 descriptors) using Schrodinger, TopoCluj, Polly, and Triplet software.
• Quantitative Structure-Activity Relationship (QSAR) formulation using a two-deep, repeated external validation method.
• Application of random forest variable importance for feature selection.

Main Results:

• Both descriptor sets, individually and combined, yielded QSAR models with reasonable prediction accuracy for BBB entry.
• A variable selection approach, utilizing the top 5% of random forest predictors, resulted in a more accurate model than using all descriptors.
• Identification of key molecular structural features influencing chemical BBB penetration.

Conclusions:

• Molecular descriptors are effective in predicting chemical BBB entry.
• Variable selection significantly enhances QSAR model performance for BBB penetration prediction.
• The identified influential descriptors provide insights into the structural determinants of BBB permeability.