stangene

Gene identifier harmonization for single-cell transcriptomics datasets.

stangene maps gene features from individual datasets into a shared canonical gene identity system using a tiered matching cascade. It preserves all original information, tracks mapping provenance, and never silently forces ambiguous mappings.

User Guide

• Quick Start
  • Installation
  • Step 1: Build reference data
  • Step 2: Harmonize a dataset
  • Step 3: Review outputs
  • Example output
• How It Works
  • Pipeline Overview
  • Feature Classification
  • Matching Cascade
  • Design Principles
• Input Formats
  • h5ad (AnnData)
  • TSV / CSV
• Reference Data
  • Building references
  • Human (HGNC)
  • Mouse (MGI + Ensembl BioMart)
  • Internal format
  • Custom reference directory
  • Versioning references

API Reference

• stangene (top-level)
  • Pipeline
  • Re-exported functions
• stangene.io
  • load_features()
  • write_results()
• stangene.classify
  • classify_features()
• stangene.references
  • ReferenceNotFoundError
  • build_reference()
  • load_reference()
• stangene.harmonize
  • HarmonizationResult
  • harmonize()
• stangene.merge
  • MergeResult
  • merge_features()
• stangene.report
  • summary()
  • conflict_report()
  • generate_markdown_report()
  • write_reports()
• stangene.species
  • SpeciesConfig
  • get_species_config()

Advanced

• Output Schema
  • Harmonization table
  • Mapping status values
  • Confidence levels
  • Summary JSON
  • Conflict report
  • Markdown report
• Conflicts and Merge Policy
  • Understanding conflicts
  • Conservative merge
  • Pitfalls to avoid
• Adding a New Species
  • Step 1: Add a SpeciesConfig
  • Step 2: Implement a reference builder
  • Step 3: Register the builder
  • Key requirements
  • Testing
• Command Line Interface
  • harmonize
  • build-refs
• Changelog
  • v0.1.0 (2026-03-24)