Genome-AI-Pathway-Designer

An AI-powered Metabolic Engineering Assistant designed to map biochemical pathways, extract enzymatic data, and optimize protein selection for synthetic biology. This tool leverages Neo4j Knowledge Graphs, ESM-2 Transformer Embeddings, and LLMs to provide detailed blueprints for producing high-value compounds like Astaxanthin.

---

Features

• **Metabolic Pathfinding: Automatically identifies the "Shortest Path" between a precursor (e.g., beta-Carotene) and a target molecule (e.g., Astaxanthin).
• **Enzyme Extraction: Integrates with KEGG and UniProt to retrieve specific EC numbers and amino acid sequences for each reaction step.
• **Transformer Embeddings: Uses Meta's facebook/esm2_t6_8M_UR50D model to convert protein sequences into 320-dimensional vectors for functional similarity search.
• **Graph-Based Intelligence: Stores complex biological relationships in Neo4j, allowing for the detection of "metabolic leaks" and co-factor requirements.
• **Genome-Host Alignment: Evaluates enzyme compatibility based on source organism data to ensure pathways are stable within a specific microbial host.
• **Automated Ingestion: Streamlined pipeline for crawling KEGG reaction maps and enriching the graph with real-time protein data.

---

Architecture

• **Knowledge Graph (Neo4j): Acts as the "Brain," storing nodes for Compounds, Reactions, and Enzymes, and edges representing metabolic transformations.
• **Protein Language Model (ESM-2): Applies a Transformer architecture to understand the "biological grammar" of enzymes and generate functional embeddings.
• **Bio-Informatics Extractor: A custom ingestion engine that bridges the gap between raw web-based databases (KEGG/UniProt) and a structured graph.
• **Vector Search: Enables mathematical comparison of enzymes to find the most efficient catalysts across different genomes.

---

Getting Started

Prerequisites

• Python 3.11+
• Docker (optional, for containerized deployment)
• Access to OpenAI API or other LLM providers like Ollama
• Vector database setup (e.g., ChromaDB)

Installation

1. Clone the repository:

   git clone https://github.com/SilasPenda/Policy-Compliance-Agent
   cd policy-compliance-auditor
   

2. Create & activate virtual environment:

   python -m venv .venv
   source .venv/bin/activate (Linux & Mac)
   ./.venv/Scripts/activate (Windows)
   

3. Install requirements:

   python -m pip install --upgrade pip
   pip install -r requirements.txt
   

4. Create a .env file and add your credentials:

5. Ingest biochemical data and generate AI embeddings:

      python ingestion/graph_ingestor.py
   

6. Launch API

   uvicorn deployment.api:app --reload
   

7. Start App

   streamlit run deployment/app.py