One specification. Consistent scientific applications.

The question

How can scientific tools reach multiple user-interface platforms without maintaining a separate implementation for every tool?

Building a separate GUI for every application-platform pair duplicates work and makes synchronized changes difficult. AutoCLIP treats an application description as a reusable specification.

Platform adapters generate interfaces while keeping application state, execution logic, post-analysis, and molecular visualization modular.

The approach

1. 01

   Describe the application

   Combine interface layout, command-line metadata, and exposed analysis parameters.

2. 02

   Validate the specification

   Convert the description into a platform-independent JSON representation and validate its structure.

3. 03

   Generate and deploy

   Use reusable adapters to create applications for PyMOL, VMD, and the web.

What I developed

• Architected a specification-driven framework with Model–View–Presenter separation.
• Implemented JSON Schema validation, Qt interface parsing, and platform-specific generators.
• Built cross-platform tests and CI/CD for Linux and Intel/Apple Silicon macOS.
• Added reproducible Pixi/Conda packaging and authenticated web deployment.

• Python
• JSON Schema
• PyMOL
• VMD
• Web

Results & validation

• Deployed nine structural-biology applications across PyMOL, VMD, and web environments.
• The architecture uses one specification per application and reusable platform adapters rather than independent implementations for every application-platform pair.

Related publications