Team

Why this project

Molstar is a powerful viewer, but its full configuration surface (representations, selections, colors, camera) is hard to discover. A chat layer that compiles user intent into MolViewSpec would make the tool more accessible — and a natural fit for AI-assisted exploration of molecular structures.

What a team could build in one day

A chatbot front-end (hosted on AWS, backed by Qwen, Ollama, or an existing Claude / OpenAI account) that:

• Accepts plain-language requests like “show the binding pocket as surface, color by hydrophobicity, hide waters.”
• Generates a corresponding MolViewSpec configuration.
• Loads it into a live Molstar viewer.

Minimum viable demo: a working round-trip where a user describes a scene and Molstar reflects the requested view.

Stretch directions

• Extend to MolViewStories for interactive, multi-step narratives.
• Suggest improvements (“would you like to also highlight the active site?”) based on the loaded structure.
• Library of saved prompt → scene templates for teaching.