GenoScope Pro is an AI-powered bioinformatics dashboard that predicts the pathogenicity of protein missense mutations in real-time. Combining Meta AI’s ESM2 transformer, AlphaFold 3D structures, and ClinVar clinical evidence to assess mutation impact instantly.
GenoScope Pro won the “Spark of Innovation” 🏆 at the BioSpark Hackathon 2025, held at D Y Patil International University.
- Powered by Meta’s ESM2 (
esm2_t6_8M_UR50D) model. - Computes mutation pathogenicity using evolutionary log-likelihood.
- Requires no labeled training data.
- Fetches PDB models directly from the AlphaFold DB.
- Uses Py3Dmol for 3D interactive rendering.
- Highlights mutation site in red for clarity.
- Queries ClinVar via MyVariant.info.
- Implements robust fuzzy matching to handle naming inconsistencies.
- Displays known pathogenicity classifications (Benign, Pathogenic, etc.)
| Component | Technology |
|---|---|
| Frontend | Streamlit |
| AI Engine | PyTorch + Hugging Face Transformers |
| Model | Meta AI ESM2 (facebook/esm2_t6_8M_UR50D) |
| Bioinformatics | BioPython, Py3Dmol |
| APIs | UniProt, AlphaFold DB, MyVariant.info |
| Language | Python 3.13 |
git clone https://github.com/Sparshr04/GenoScope-Pro.git
cd GenoScope-Pro/Note: This project supports uv, a new fast Python package manager.
If you don’t have uv: pip install uv
#If you use uv, you can install and run the app instantly:
uv venv
source .venv/bin/activate # Windows: .venv\Scripts\activate
# installs dependencies from pyproject.toml
uv sync
#Run the app
uv run streamlit run app.pypython -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
pip install -r requirements.txt
#Run the app
streamlit run app.pygenoscope-pro/
├── app.py # Streamlit Interface
├── assets
│ ├── demo3.gif
│ └── flow.jpeg
├── config.py # Configuration and constants
├── core
│ ├── ai_engine.py. # ESM2 Zero-shot Inference Logic
│ └── data_client.py # Handles UniProt, AlphaFold, ClinVar APIs
├── dataset.py
├── examples
│ └── mutations_list.txt
├── requirements.txt # Dependencies
├── pyproject.toml
├── README.md
├── ui
│ └── visualizer.py # 3D Py3Dmol Visualization
If you are unsure about the UniProt IDs and Mutation Pairs. You can use the examples pairs. But all pairs might not work, although you can cross match the mutations and proteins! Just copy-paste the content from
examples/mutations_list.txtfile