How rdk/p2rank works
Key Interactions and Data Flow
• Data flow from input files through processing stages
• Configuration parameters affecting each component
• Model loading and application paths
• Output generation flows
Detailed Component Breakdown
a) Feature Extraction Engine:
• Surface Point Generation
• Chemical Feature Calculators
• Conservation Feature Processors
• Structural Feature Analyzers
b) ML Prediction Engine:
• Model Loading/Application
• Score Transformation
• Probability Calibration
c) Output Generation:
• CSV Results Generator
• Visualization Script Generator
• SAS Points Calculator
External Dependencies (show as separate connected components)
• Java Runtime (17-23)
• PyMOL/ChimeraX Integration
• FastRandomForest
• BioJava
• CDK (Chemistry Development Kit)
Specific elements
a) File Formats (input):
• PDB (.pdb)
• mmCIF (.cif)
• BinaryCIF (.bcif)
• Compressed formats (.gz, .zst)
b) File Formats (output):
• Predictions CSV
• Residues CSV
• Visualization scripts (.pml, .cxc)
• SAS points data
Special Features
• Parallel Processing Capabilities
• Rescoring Pipeline
• Model Configuration System
• Dataset Processing Framework
Use the following visual conventions:
• Rectangles: Main processing components
• Hexagons: External systems/dependencies
• Clear labels for all components