How biocypher/graphpack works
Project Type and Purpose
• GraphPack is a Python library and CLI tool for compressing and visualizing large‐scale graphs (e.g., PPI networks).
• It exposes both a command‐line interface and a programmatic API, implements multiple compression algorithms, produces mapping files for lossless compression, and renders interactive visualizations.
Key Architectural Components
a. Entry Points
• CLI script (installed via setuptools entry_point)
• Python API (importable graphpack package)
b. Core Modules
• src/graphpack/compression.py ← implements compression algorithms
• src/graphpack/utils.py ← I/O, graph loading/saving, mapping file generation
c. Configuration & Assets
• src/graphpack/config ← JSON/JS files holding physics and style options for visualization
d. Demo & Examples
• src/graphpack/demo/demo.py & sankey.py ← example workflows combining compression & visualization
e. Documentation
f. Tests
• tests/… ← unit tests for compression and utils
g. Data
• data/input/*.txt ← sample graph inputs for demos/tests
External Dependencies
• Network manipulation: NetworkX (or igraph)
• Numerical & ML: NumPy, SciPy
• Visualization: D3.js (embedded via config JS), Matplotlib or Bokeh for static plots
• CLI: Click or argparse
• Documentation: Sphinx
Relationships and Data Flows
• CLI/API → invokes compression module
• Compression module ↔ utils (for graph I/O & mapping file production)
• Compression output → passed to visualization sub‐module along with config assets
• Visualization sub‐module → emits HTML/JSON/SVG outputs
• Demo scripts → orchestrate CLI/API calls to produce end‐to‐end examples
• Tests → call core modules to validate algorithm correctness
Architectural Patterns & Design Principles
• Layered architecture:
1. Presentation (CLI/UI)
2. Application (API entry‐points, demos)
3. Domain/Business logic (compression algorithms)
4. Infrastructure (utilities, I/O, config files)
• Modular plugin‐style for adding new algorithms—each algorithm implemented as a function/class in compression.py.
• Separation of concerns: compression vs. visualization vs. I/O vs. configuration.
Final Advice
• Major directories (src, docs, tests, data) as contextual side‐bars.
• Pluggable algorithm interface in the compression module.
• How config assets drive the visualization options.