B3DB

B3DB compiles curated molecular data to support prediction of blood–brain barrier (BBB) permeability using numerical log BB values and categorical BBB+ and BBB- labels for applications in CNS drug development and related research.

Key Features:

• Extensive Dataset: Compiles data from 50 published resources into a large benchmark dataset to increase chemical diversity compared with smaller prior studies.
• Categorization by Experimental Uncertainty: Assigns categories based on experimental uncertainty to convey confidence levels associated with each entry.
• Numerical and Categorical Data: Provides numerical log BB values for 1,058 compounds and categorical BBB+ or BBB- labels for 7,807 compounds.
• Physicochemical Properties: Includes detailed physicochemical properties for molecules to enable analysis of characteristics that distinguish BBB+ and BBB- compounds.

Scientific Applications:

• Lead Discovery in CNS Drug Development: Predicts which compounds can cross the BBB to aid identification of potential CNS drug leads.
• Machine Learning Integration: Serves as training and benchmarking data for machine learning and traditional statistical models to predict BBB permeability.

Methodology:

Data were meticulously curated from multiple published resources and entries were categorized by experimental uncertainty; the dataset is provided for use with traditional statistical methods and machine learning techniques.