Alignment: Align representations using guidance graph
Joint Training: Optimize reconstruction and alignment simultaneously
Reference Files
This skill includes comprehensive documentation in references/:
api_models.md - API Reference
Pages: 48
Complete API documentation for all public functions and classes
Model classes: SCGLUEModel, PairedSCGLUEModel, SCCLUEModel
Neural network modules and utilities in scglue.models.nn
Plugin system for training extensions
Probabilistic model registration and configuration
Key sections:
Model fitting with fit_SCGLUE()
Base classes for custom model development
Data encoders/decoders for different data types
Training plugins and callbacks
data_management.md - Data Processing & Integration
Pages: 25
Comprehensive data preprocessing workflows
Guidance graph construction methods
Metacell-based correlation analysis
Partially paired dataset handling
Example datasets and case studies
Key sections:
Stage 1 preprocessing pipeline (RNA + ATAC)
Genomic coordinate handling and annotation
Custom guidance graph construction
Paired cell identification via obs_names
getting_started.md - Installation & Tutorials
Pages: 3
Installation instructions (conda/pip)
Complete preprocessing tutorial with SNARE-seq data
Step-by-step guidance graph construction
Model training and evaluation workflows
Key sections:
Environment setup and optional dependencies
End-to-end integration pipeline
Data visualization and quality control
Working with This Skill
For Beginners
Start with getting_started.md for:
Installation and environment setup
Basic data preprocessing concepts
Simple integration workflows
Understanding AnnData and NetworkX structures
Recommended workflow:
Read the installation guide and set up environment
Follow the complete preprocessing tutorial
Try the basic GLUE model training example
Explore embedding extraction and visualization
For Intermediate Users
Use data_management.md for:
Advanced preprocessing techniques
Custom guidance graph construction
Working with partially paired datasets
Metacell analysis and correlation methods
Common tasks:
Integrating custom multi-omics datasets
Building domain-specific guidance graphs
Optimizing model parameters for specific data types
Implementing quality control metrics
For Advanced Users
Reference api_models.md for:
Custom model architecture development
Extending the framework with new probabilistic models
Implementing custom training plugins
Advanced neural network module design
Advanced applications:
Developing new encoders/decoders for novel data types
Creating custom loss functions and training strategies
Integrating external knowledge sources
Scaling to large multi-modal datasets
Navigation Tips
Use view command to read specific reference sections
Search for function names using grep in reference files
Code examples include proper syntax highlighting
All examples are extracted from official documentation
Resources
references/
Organized documentation extracted from official sources:
Detailed explanations of all scGLUE concepts and methods
Code examples with language annotations and syntax highlighting
Links to original documentation for further reading
Structured table of contents for quick navigation
scripts/
Add helper scripts here for:
Automated preprocessing pipelines
Custom guidance graph construction
Batch model training and evaluation
Integration quality assessment
assets/
Store templates and examples:
Configuration file templates
Example datasets in proper format
Visualization templates
Best practice checklists
Notes
Documentation Coverage: 100% coverage of official scGLUE documentation (76 pages across 3 main sections)
Real Examples: All code examples extracted from actual tutorials and API documentation
Practical Focus: Emphasis on actionable workflows and common use cases
Multi-level Support: Guidance available for beginners through advanced users
Quality Assurance: All examples tested against official documentation standards
Updating
To refresh this skill with updated documentation:
Re-run the documentation scraper with the same configuration
The skill will be rebuilt with the latest information from scGLUE official docs
All reference files will be updated while preserving skill structure
Installation Prerequisites
Before using this skill, ensure you have scGLUE installed:
# Via conda (recommended)
conda install -c conda-forge -c bioconda scglue # CPU only
conda install -c conda-forge -c bioconda scglue pytorch-gpu # With GPU
# Via pip
pip install scglue
# Optional: faiss for speedup with metacell aggregation
# Follow official faiss installation guide
Common Troubleshooting
Memory Issues: Reduce dataset size or use metacell aggregation
GPU Errors: Install pytorch-gpu version and check CUDA compatibility
Graph Construction: Ensure proper genomic coordinates and edge attributes
Model Convergence: Check learning rate settings and data preprocessing quality