Custom BioChatter Light: Simple Use Case

For prototyping your own text2cypher workflow, it can be useful to have a simple interface for troubleshooting your queries. This vignette demonstrates how to customise the pre-built BioChatter Light Docker compose to turn off all tabs except the Knowledge Graph (KG) tab. This allows the deployment of an integrated KG build, deployment, and web app for LLM-based question answering.

Build your KG

First, build your KG. For this example, we use the BioCypher Pole KG as a demo KG. The KG is based on an open-source dataset of crime statistics in Manchester. The schema of the demo KG is described in the Knowledge Graph RAG vignette. For building your own KG, refer to the BioCypher documentation.

This KG is built, imported, and deployed in the first three stages of the docker-compose.yml file.

Configure the BioChatter Light Docker container

We provide a simple way to customise the BioChatter Light Docker container to show only select components. We can provide these settings via environment variables, so in the case of running from docker-compose.yml, we can set these in the environment section of the app service.

services:
  ## ... build, import, and deploy the KG ...
  app:
    image: biocypher/biochatter-light:0.6.10
    container_name: app
    ports:
      - "8501:8501"
    networks:
      - biochatter
    depends_on:
      import:
        condition: service_completed_successfully
    environment:
      - OPENAI_API_KEY=${OPENAI_API_KEY}
      - DOCKER_COMPOSE=true
      - CHAT_TAB=false
      - PROMPT_ENGINEERING_TAB=false
      - RAG_TAB=false
      - CORRECTING_AGENT_TAB=false
      - KNOWLEDGE_GRAPH_TAB=true

In this example, we provide our OpenAI API key and set the DOCKER_COMPOSE flag to true, which tells BioChatter Light to connect to the KG on the Docker network, which uses the service name as the hostname, so in this case, deploy instead of the default localhost.

Authentication

For using the app with the standard OpenAI LLM, we need to provide the OPENAI_API_KEY environment variable. This key can be obtained from the OpenAI website.

We then turn off all default tabs (chatting, prompt engineering, RAG, and the correcting agent) and turn on the KG tab. Running the docker compose with these settings will build and deploy the KG and the BioChatter Light web app with only the KG tab enabled.

git clone https://github.com/biocypher/pole
cd pole
docker-compose up -d

We are constantly expanding our repertoire of BioChatter Light tabs, so check back for more options in the future. Creating your own tabs is also accessible via our modular architecture and the simple Streamlit framework for UI design. Check the advanced vignette and the codebase of BioChatter Light for more information.

Open-source LLMs

You can use a locally hosted open-source LLM model by using the docker-compose-ollama.yml as an example (this deployment requires Ollama to be installed on your machine). See more here.