BioChatter is a new open source Python package (v0.11.1 released June 6 2025) designed to bring conversational AI into biomedical research workflows.

This library provides a clean backend interface for integrating large language models with biological knowledge graphs via BioCypher and common biomedical tasks. It abstracts away provider specific details with a unified API for sending prompts, retrieving structured outputs, and handling model completion logic over multiple LLMs.

Getting started

Install the package via pip:

pip install biochatter

Then initialize and call a model:

from biochatter import BioChatter

bc = BioChatter(provider="openai", api_key="YOUR_KEY")
response = bc.ask("List key genes involved in oxidative stress response")
print(response)

BioChatter converts your question into a structured query, invokes an LLM, and formats the response as JSON or a DataFrame. It also supports:

• integration with BioCypher for knowledge graph grounding
• evaluation metrics such as accuracy and F1 for test sets
• model agnostic chaining to compare GPT 4, LLaMA, and others

Why it matters

BioChatter streamlines building apps like gene disease extractors, literature summarizers, or dialog agents for molecular data without scaffolding each component manually. The built in evaluation framework lowers the barrier for reproducible benchmarks in biomedical NLP.

Researchers working on hypothesis generation, clinical extractive tasks, or biomedical QA now have an easy way to prototype and compare conversational AI pipelines.

Next steps

Explore documentation and examples at the GitHub repo. Start experimenting by connecting to your lab’s knowledge graph or loading custom biomedical prompts. Share your use cases and help grow the community.

BioChatter lays the foundation for more conversational and data driven tools in biology, offering a versatile core for integrating LLMs thoughtfully into scientific workflows.

Source

https://pypi.org/project/biochatter/