A machine learning–driven system that predicts enzyme efficiency using deep neural networks and accelerates metabolic pathway optimization through intelligent search and explainable AI.
This project bridges Synthetic Biology, Machine Learning, and Generative AI techniques to reduce experimental cost and improve biological decision-making.
In synthetic biology, identifying efficient enzymes for metabolic pathways is crucial but experimentally expensive and time-consuming.
This project uses Deep Learning models trained on enzyme–substrate interaction data to accurately predict enzyme efficiency and applies graph search algorithms to optimize pathways.
Key goals:
- Predict enzyme efficiency with high accuracy
- Optimize metabolic pathways programmatically
- Provide interpretability for biological trust
- Deploy as a scalable web/API system
- 🧠 Deep Neural Network (DNN) for enzyme efficiency prediction
- 📊 Trained on 10,000+ enzyme–substrate samples
- ⚡ 92% prediction accuracy
- 🧭 A* Search–based metabolic pathway optimization (2.3× speedup)
- 🔍 Explainable AI (SHAP & LIME) for biological feature interpretation
- 🌐 Flask-based Web Application
- 🐳 Dockerized REST API (Kubernetes-ready)
- 📉 Reduced manual biological validation time by 40%
- Python 3.x
- Flask
- TensorFlow / Keras
- Scikit-learn
- NumPy
- Pandas
- SHAP
- LIME
- A* Search
- Graph-based pathway modeling
- Docker
- REST APIs
- Kubernetes (deployment-ready)
- HTML5
- CSS3
- Jinja2 Templates