CROssBAR-LLM

This repo is created within the scope of the CROssBARv2 project to convert natural language questions into Cypher query language. We leverage LangChain to construct Large Language Model (LLM) chains for the generation of Cypher queries and the subsequent parsing of the resulting output. Streamlit is employed to create an interactive user interface for this process.

Installation

poetry

First, ensure that you have poetry installed on your system.

poetry --version

If it's not installed, install via pip:

pip install -U poetry

To install the package:

poetry install

To get into the poetry virtual env:

If poetry version <2 :

poetry shell

If poetry version 2>= :

poetry env activate

pip

pip install -r requirements.txt

Usage

Setting up Environment Variables

To use this package, ensure that you have .env file present in the root folder of it. You should use the template below:

# Environment: 'development' (default) or 'production'
# See ENVIRONMENT.md for more details
CROSSBAR_ENV=development

OPENAI_API_KEY="sk-************************************************"
GEMINI_API_KEY="AI*****************-**-******-*********"
ANTHROPIC_API_KEY="sk-ant-***-***********************-****"
NVIDIA_API_KEY="nvapi-************************************"
GROQ_API_KEY="gsk_****************************************"
REPLICATE_API_KEY="***************************************"
NEO4J_USERNAME="user"
MY_NEO4J_PASSWORD="password"
NEO4J_DATABASE_NAME="neo4j"
NEO4J_URI="neo4j://localhost:7687"

Running the Application

The CROssBAR-LLM application consists of two main components:

• Backend: FastAPI server that handles natural language to Cypher query conversion
• Frontend: React application providing the user interface

Option 1: Running in Development Mode (Recommended for development)

Backend Setup:

1. Navigate to the backend directory:

   cd crossbar_llm/backend

2. Run the FastAPI server:

   # If using poetry
   poetry run uvicorn main:app --reload --host 0.0.0.0 --port 8000
   
   # If using pip
   uvicorn main:app --reload --host 0.0.0.0 --port 8000

   The backend API will be available at http://localhost:8000

Frontend Setup:

1. In a new terminal, navigate to the frontend directory:

   cd crossbar_llm/frontend

2. Install dependencies:

   npm install
   # or if using pnpm
   pnpm install

3. Start the development server:

   npm start
   # or if using pnpm
   pnpm start

   The frontend will be available at http://localhost:3000

Option 2: Running with Docker (Recommended for production)

1. Build the Docker image:

   docker build -t crossbar-llm .

2. Run the container:

   docker run -p 8000:8000 -p 8501:8501 --env-file .env crossbar-llm

   • Backend API: http://localhost:8000
   • Frontend: http://localhost:8501

Option 3: Running with the Startup Script

If you have all dependencies installed locally, you can use the provided startup script:

./startup.sh

This will start both the backend (port 8000) and frontend (port 8501) services.

Using the Application

1. Open your web browser and navigate to:

   • Development mode: http://localhost:3000
   • Production/Docker mode: http://localhost:8501

2. In the interface, you can:

   • Enter natural language questions about the CROssBAR knowledge graph
   • View the generated Cypher queries
   • See query results and visualizations
   • Explore the knowledge graph relationships

3. Example queries you can try:

   • "Show me all proteins related to cancer"
   • "What drugs target EGFR?"
   • "Find pathways associated with diabetes"

API Documentation

When the backend is running, you can access:

• Interactive API documentation: http://localhost:8000/docs
• Alternative API documentation: http://localhost:8000/redoc

Troubleshooting

• Ensure Neo4j database is running and accessible at the URI specified in .env
• Check that all required API keys are properly set in the .env file
• For CORS issues in development, ensure CROSSBAR_ENV=development is set
• If ports 8000 or 8501 are already in use, modify the port numbers in the respective commands

Environment Configuration

The application supports both development and production environments with different security settings:

• Development mode (default): Disables CSRF protection and rate limiting for easier local development
• Production mode: Enables full security features

For details on configuring the environment, see ENVIRONMENT.md.

This repo is currently under development. Therefore, you may encounter some problems while replicating this repo. Feel free to open issue about it.