This project uses RDKit to extract molecular descriptors from drug SMILES notation and applies machine learning techniques to predict Blood Brain Barrier (BBB) permeability.

• Molecular Analysis: Extract 200+ molecular descriptors from SMILES notation
• Similarity Analysis: Compare molecular similarity using Morgan fingerprints and Tanimoto coefficients
• Exploratory Data Analysis: Comprehensive EDA with PCA visualization
• Machine Learning Models: Implemented Random Forest, SVM, and Logistic Regression for BBB prediction
• Model Evaluation: Cross-validation, confusion matrices, ROC curves, and feature importance analysis

BBB_permeability_prediction/
├── README.md                 # This file
├── requirements.txt          # Python dependencies
├── .gitignore               # Git ignore file
├── LICENSE                  # MIT License
├── data/                    # Data directory
│   ├── sample_data.csv      # Sample dataset
│   └── README.md           # Data documentation
├── src/                     # Source code
│   ├── bbb.py              # Main analysis script
│   ├── ml_models.py        # Machine learning models
│   └── predict_bbb.py      # Prediction script
├── notebooks/               # Jupyter notebooks
│   └── bbb_analysis.ipynb  # Interactive analysis
├── docs/                    # Documentation
│   └── molecular_descriptors.md
├── results/                 # Output files
│   ├── plots/              # Generated plots
│   └── models/             # Trained models
└── tests/                  # Unit tests
    └── test_bbb.py

• Python 3.7 or higher
• pip package manager

1. Clone the repository:

git clone <repository-url>
cd BBB_permeability_prediction

2. Create a virtual environment (recommended):

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

1. Place your BBB dataset as data/BBB_datasets.csv with columns:

   • SMILES: Chemical structure in SMILES notation
   • Class: BBB permeability class (BBB+ or BBB-)

2. Run the analysis:

python src/bbb.py

3. Predict BBB permeability for new molecules:

python src/predict_bbb.py "CC(=O)NC1=CC=C(C=C1)O" "CN1C=NC2=C1C(=O)N(C(=O)N2C)C"

Or using Makefile:

make predict SMILES="CC(=O)NC1=CC=C(C=C1)O"

For interactive analysis, use the Jupyter notebook:

jupyter notebook notebooks/bbb_analysis.ipynb

The script expects a CSV file with the following columns:

The script extracts 200+ molecular descriptors including:

• Molecular weight
• LogP (lipophilicity)
• Number of rotatable bonds
• Hydrogen bond donors/acceptors
• Topological descriptors
• And many more...

The script generates:

• Molecular similarity analysis
• PCA visualization of molecular descriptors
• Statistical summaries of the dataset
• Plots showing drug clustering by BBB permeability
• Machine learning model training and evaluation
• Feature importance analysis
• Model performance comparison
• Confusion matrices and ROC curves
• Trained models saved for future predictions

• Paracetamol: CC(=O)NC1=CC=C(C=C1)O
• Caffeine: CN1C=NC2=C1C(=O)N(C(=O)N2C)C
• Theophylline: CN1C2=C(C(=O)N(C1=O)C)NC=N2
• MDMA: CC(CC1=CC2=C(C=C1)OCO2)NC

• RDKit: Cheminformatics toolkit
• pandas: Data manipulation
• numpy: Numerical computing
• matplotlib: Plotting
• seaborn: Statistical visualization
• scikit-learn: Machine learning

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

This project is licensed under the MIT License - see the LICENSE file for details.

• RDKit community for the excellent cheminformatics toolkit
• Original Colab notebook: BBB Analysis

The project implements three machine learning algorithms:

• Advantages: Handles non-linear relationships, provides feature importance
• Parameters: 100 estimators, max depth 10, min samples split 5
• Use Case: Best for interpretable predictions with feature importance

• Advantages: Effective for high-dimensional data, good generalization
• Parameters: RBF kernel, C=1.0, gamma='scale'
• Use Case: Good performance on molecular descriptor data

• Advantages: Fast training, interpretable coefficients
• Parameters: C=1.0, max iterations 1000
• Use Case: Baseline model and fast predictions

• Cross-validation: 5-fold CV for robust performance estimation
• Metrics: Accuracy, precision, recall, F1-score, AUC-ROC
• Feature Importance: Top 20 most important molecular descriptors
• Visualization: Confusion matrices, ROC curves, performance comparison

• Implement machine learning models (Random Forest, SVM, Logistic Regression)
• Add feature importance analysis
• Cross-validation and model evaluation
• Hyperparameter tuning with GridSearch
• Neural Networks (Deep Learning)
• Web interface for drug prediction
• API for batch processing
• Integration with drug databases

1. RDKit installation issues: Try using conda instead of pip:

   conda install -c conda-forge rdkit

2. Missing dataset: Ensure BBB_datasets.csv is in the data/ directory

3. Memory issues: For large datasets, consider processing in batches

• Check the Issues page
• Create a new issue with detailed error information
• Include your Python version and operating system

If you use this project in your research, please cite:

@software{bbb_prediction,
  title={Blood-Brain Barrier Permeability Prediction},
  author={Your Name},
  year={2024},
  url={https://github.com/yourusername/BBB_permeability_prediction}
}