AI in drug discovery for CNS: blood-brain barrier permeability prediction

This data app predicts whether a chemical compound can cross the blood-brain barrier (BBB) using data science and cheminformatics techniques. It demonstrates how an AI implementation can autonomously analyze and propose candidate molecules for central nervous system (CNS) drug discovery, where users query the system with chemical structures and explore similar compounds, supporting CNS drug discovery.

User provides a chemical structure as a SMILES string. (A SMILES string is a text-based notation that encodes a molecule’s structure, widely used in computational and cheminformatics applications.)

The app finds similar compounds in a large scientific database (B3DB) and assesses their BBB permeability.

Users can filter results by molecular weight or predicted BBB class, and visualize structures interactively.

The data comes from the Blood-Brain Barrier Database (B3DB), which was obtained by downloading curated experimental datasets from the official B3DB GitHub repository and merging multiple source files into a single, cleaned dataset for analysis.

Usage instructions:

Enter a SMILES string into the text box (e.g., "CCO" for ethanol, "CC(=O)O" for acetic acid).

Select how many results you wish to see.

SMILES input

Number of molecules to visualize

Minimum molecular weight: Lower bound for filtering displayed compounds

Maximum molecular weight: Upper bound for filtering displayed compounds

Filter by BBB class to see only those predicted to cross or not cross the blood-brain barrier.

Minimum molecular weight

Maximum molecular weight

compounds predicted to cross (or not cross) the blood-brain barrier

BBB+

.css-15w88e5{color:var(--chakra-colors-fg-neutral-primary);font-weight:inherit;letter-spacing:-0.09px;}AI in drug discovery for CNS: blood-brain barrier permeability prediction

Usage instructions:

Results

AI in drug discovery for CNS: blood-brain barrier permeability prediction