Streamlit interface for ChatMol’s Agentic Molecular Visualization (AMV).
This app runs a Streamlit GUI that lets you interact with an LLM-powered PyMOL agent. The agent can:
streamlit (tested with 1.35.0+)openai Python SDKchatmol package at hereYou can install the main dependencies with conda/pip:
conda install -c conda-forge pymol-open-source
pip install streamlit==1.35.0 openai chatmol
The app expects one or more API key environment variables. By default the UI is configured to use OPENAI_API_KEY for all models. Set it before starting Streamlit:
export OPENAI_API_KEY="sk-..."
If you prefer different env var names for different model roles (PyMOL commands, view analysis, agent) you can change them in the sidebar at runtime — the app lets you specify the name of the environment variable to read for each role.
From the chatmol-streamlit directory run:
streamlit run app.py
The Streamlit sidebar contains a Model & API settings section where you can:
PyMOL commands, View analysis, and the Agent.https://api.openai.com/v1).Changes are applied immediately for new requests.
Try the test prompt shown in the sidebar (also used in the app):
Highlight aromatic sidechains of 1pga and prepare to scientific publication level.
When you submit a prompt the app will display assistant messages in the chat area and show the latest screenshot.png produced by PyMOL.
OPENAI_API_KEY).pymol-open-source is installed and that chatmol’s start_pymol_gui() works on your platform.screenshot.png in the working directory — check permissions if saving fails.