User Design
User lay out the input, output, and optional update areas in Qt Designer, then select the CLI and post-analysis parameters that should be exposed.
All projects
Research software
One specification, multiple scientific interfaces
Automatic plugin generation for VMD, PyMOL, and the web
9
bioinformatics application
1
formal specification per application
3
target environments
Context
The problem
Scientific command-line applications are powerful but expensive to expose through multiple user-interface ecosystems. A conventional approach creates a separate implementation for every application-platform pair, which duplicates logic and makes synchronized maintenance difficult.
The framework treats the interface design as a reusable specification rather than platform-specific source code. It follows a Model–View–Presenter separation so that application state, interface rendering, execution logic, post-analysis, and molecular visualization remain modular.
Method
How the system works
Auto Formalize
Automatically convert the layout and metadata into a validated JSON representation containing widget types, geometry, CLI semantics, outputs, and update behavior.
Auto Generate
Use platform adapters to synthesize native applications for Tkinter/VMD, pymol.Qt/PyMOL, and Panel with NGL or Three.js.
Role
My contribution
- Architected the specification-driven workflow and Model–View–Presenter boundaries.
- Implemented JSON schema validation, Qt UI parsing, CLI metadata integration, and platform-specific code generators.
- Built communication layers for VMD sockets, the PyMOL command API, and embedded NGL/Three.js web viewers.
- Created canonical generated-project structures, examples, documentation, and end-to-end generation commands.
Result
Outcomes and scientific value
✓
Reduces the development structure from application × platform implementations to one specification per application plus one reusable generator per platform.
✓
Provides a single source of truth for interface changes and keeps desktop and web versions synchronized.
✓
Makes structural-bioinformatics CLI tools accessible without coupling scientific logic to a particular GUI toolkit.
Resources
Papers, code, and documentation
Related work
Publications
2026
Preprint · arXiv · arXiv:2602.16047
A Unified, Cross-Platform Framework for Automatic GUI and Plugin Generation in Structural Bioinformatics and Beyond
Contribution: Design, implementation, platform generators, examples, and writing manuscript.