PaperToData

An Agentic RAG System for Vaccines, complete with frontend and agentic backend.

VaxChat utilizes a variety of tools to answer user queries using retrieved data.

Table of Contents

• Installation
• Usage
• Example

Installation

1. Clone the repository:

> git clone https://github.com/crazab0i/VaxChat.git

2. Install dependencies:

> pip install -r requirements.txt

3. Set up .env files within the project root:

   • langchain.env
   • llm.env
   • email.env
   • neo4j.env

   Make sure that the fields are encapsulated as strings:

   FIELD_NAME="value_here"

   langchain.env

   LANGCHAIN_ENDPOINT={YOUR_ENDPOINT}
   LANGCHAIN_TRACING="true" 			//default is true
   LANGCHAIN_API_KEY={YOUR_API_KEY}

   llm.env

   DEPLOYMENT={MODEL_NAME} 			// gpt-4o-mini is good
   API_VERSION={VERSION} 				// 2024-06-01 is good
   API_KEY={YOUR_API_KEY}
   ENDPOINT={YOUR_ENDPOINT}
   ORGANIZATION={YOUR_ORGANIZATION} 

   email.env

   EMAIL={ENTREZ_EMAIL}    //use an email for Entrez API, required for PubMed API searching

   neo4j.env

   NEO4j_URI={URI}
   NEO4J_USER={USER}
   NEO4J_PASSWORD={PASSWORD}

Usage

To run the program, you need to start the frontend and backend:

Frontend

> npm run dev               // within the frontend folder

Backend

> python3 ./vaxchat.py   // within the backend folder

If python3 does not work:

> python ./vaxchat.py

Example

Frontend

Backend Pipeline