Pharmacogenetic Prediction and Therapeutic Efficacy via Deep Learning.
Pharmagen is an advanced bioinformatics platform designed to predict patient phenotypic responses to specific drugs based on their genomic profile. It leverages deep learning models (Deep Factorization Machines - DeepFM) to analyze complex interactions between genetic variants (SNPs, haplotypes) and drug chemical structures.
The system integrates a comprehensive pipeline ranging from raw genomic file processing (VCF/FASTQ) to clinical prediction, offering robust support for decision-making in precision medicine.
- Hybrid DeepFM Architecture: Combines the memorization capability of low-order interactions with the generalization power of deep neural networks to model drug-gene relationships.
- Granular Clinical Outcomes: Predicts specific pharmacological impacts (e.g., "PK: Poor Metabolizer", "Toxicity: High Risk"), not just generic associations.
- Multi-Label Inference: Capable of identifying multiple simultaneous effects (e.g., Low Efficacy AND Toxicity) for a single drug-variant pair.
- Integrated Bioinformatics Pipeline: Tools for read cleaning (QC), alignment, and variant calling for Next-Generation Sequencing (NGS) data.
- Automated Optimization (AutoML): Native integration with Optuna for efficient model hyperparameter search.
- Interactive Web Interface: A modern Streamlit dashboard for single-patient and batch cohort analysis.
- Operating System: Linux (Recommended/Ubuntu), macOS, Windows (WSL2 recommended).
- Python: Version 3.10 (Strict).
- Hardware:
- CPU: Minimum 4 cores.
- RAM: 8GB (16GB+ recommended for training).
- GPU: NVIDIA CUDA-compatible (Optional, but highly recommended for training).
- CPU: Minimum 4 cores.
We recommend using uv for ultra-fast dependency management, though pip and conda are also supported.
# 1. Install uv if not already installed
curl -LsSf https://astral.sh/uv/install.sh | sh
# 2. Clone the repository
git clone https://github.com/Aderfi/Pharmagen.git
cd Pharmagen
# 3. Sync environment (automatically creates venv and installs deps)
uv syncgit clone https://github.com/Aderfi/Pharmagen.git
cd Pharmagen
# Create virtual environment
python3.10 -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install dependencies
pip install -e .To activate the environment with uv: uv run python main.py [args]
For standard environment: python main.py [args]
Ideal for exploring functionalities without memorizing commands.
python main.py --mode menuGenerate the master training dataset from raw PharmGKB annotations.
uv run python src/data/etl_pipeline.pyTrain the model using the processed data.
# Standard training (Single Split)
python main.py --mode train --model Phenotype_Effect_Outcome --input data/processed/final_training_data.tsv
# Robust Validation (5-Fold Cross-Validation)
python main.py --mode train --model Phenotype_Effect_Outcome --input data/processed/final_training_data.tsv --cv 5Start the Streamlit dashboard to visualize predictions.
./run_app.shFor further details, see docs/PROJECT_STRUCTURE.md.
src/: Core source code (models, pipeline, CLI).config/: TOML configuration files (paths, hyperparameters).data/: Data storage (raw, processed, dicts).reports/: Training outputs, plots, and Optuna reports.tests/: Unit and integration tests.
Contributions are welcome. Please open an issue to discuss major changes before submitting a pull request.
- Fork the project.
- Create your feature branch (
git checkout -b feature/AmazingFeature). - Commit your changes (
git commit -m 'Add some AmazingFeature'). - Push to the branch (
git push origin feature/AmazingFeature). - Open a Pull Request.
This project is licensed under the GNU GPLv3. See the LICENSE file for more details.
Author: Adrim Hamed Outmani (@Aderfi)