Query tool AI

Neurobagel is a federated data ecosystem that allows researchers and other data users to find and consume research data that has to remain at their original institute for data governance reasons.

Currently, the researcher or the data user has to answer a number of queries to get the desired results and it often requires iteration.

The aim of query-tool-ai would be to make this search process more user-friendly by adding an LLM style chatbot interface. This is to be done by utilizing large language models that will be able to interpret the user prompts and initiate the API calls accurately giving the user the desired results.

Local installation -

Clone the repository :

git clone https://github.com/neurobagel/query-tool-ai.git
cd query-tool-ai

After cloning the repository, you can choose to either use the Dockerized version or run the application locally using Python but before proceeding with either you need to set the environment variables.

Set the environment variables -

Environment Variable	Type	Required	Default Value if Not Set	Example
NB_API_QUERY_URL	string	Yes	-	https://api.neurobagel.org/query/?
HOST	string	No	0.0.0.0	127.0.0.1
PORT	integer	No	8000	8080

Option 1 : Docker :

• First, install Docker.

• Build the image using the Dockerfile

  After cloning the current repository, build the Docker image locally:

  docker build -t neurobagel-query-tool-ai .

• Run the container from the built image:

  Start the container from the built image. Ensure you map the necessary volumes and set the container name.

  docker run -d \
  -p 8000:8000 \
  -v ollama:/root/.ollama \
  -v /home/user/data:/app/output/ \
  --name query-tool-ai-container neurobagel-query-tool-ai

  Then you can access the query tool ai at http://localhost:8000

  Note: the query tool ai is listening on port 8000 inside the docker container, replace port 8000 by the port you would like to expose to the host. For example if you'd like to run the tool on port 8080 of your machine you can run the following command:

  docker run -d \
  -p 8080:8000 \
  -v ollama:/root/.ollama \
  -v /home/user/data:/app/output/ \
  --name query-tool-ai-container neurobagel-query-tool-ai

• Verify Host and Port

  To ensure the server is running on localhost, you can check the logs or try accessing the server in your browser or with curl(assuming the the query tool ai is listening on port 8000):

  curl -X GET "http://localhost:8000/"

  You should see the response:

  {"message": "Welcome to the Neurobagel Query Tool AI API"}

Option 2 : Python

• Create and activate a virtual environment:

  python3 -m venv venv
  source venv/bin/activate

• Install the required packages and pull the Ollama model:

  pip install -r requirements.txt
  ollama pull mistral

• Run the FastAPI Server:

  python3 -m app.main

  This command starts the FastAPI server on http://localhost:8000. (assuming PORT=8000)

• Verify Host and Port

  To ensure the server is running on localhost, you can check the logs or try accessing the server in your browser or with curl:

  curl -X GET "http://localhost:8000/" 

  You should see the response:

  {"message": "Welcome to the Neurobagel Query Tool AI API"}

Interacting with the Query Tool AI

After the local installation is complete, you can ask your query in the following 2 ways.

API Interaction -

You can interact with the FastAPI application by sending a POST request to the generate_url endpoint. Here’s how you can do it using curl (assuming the the query tool ai is listening on port 8000):

curl -X POST "http://localhost:8000/generate_url/" -H "Content-Type: application/json" -d '{"query": "your query here"}'

Replace "your query here" with the actual query you want to test.

Python Script Interaction (Optional) -

• If you have completed the local installation using docker, write the following command in the terminal.

  docker exec -it query-tool-ai-container python3 -m app.api.url_generator

• If you have completed the local installation using python, write the following command in the terminal.

  python3 -m app.api.url_generator

You should see the following prompt -

Enter user query (or 'exit' to quit): 

Enter your query to get the desired API URL.

Testing

Neurobagel API utilizes Pytest framework for testing.

To run the tests first make sure you're in repository's main directory.

You can then run the tests by executing the following command in your terminal:

pytest tests

License

Neurobagel API is released under the terms of the MIT License