Single-cell architectural surgery for reference atlas mapping without data sharing
scArches uses transfer learning and architectural surgery to seamlessly map new single-cell query datasets onto massive reference atlases (like the Human Cell Atlas) without retraining or sharing raw data.
Architectural surgery updating reference neural networks with query-specific parametersPreserves privacy and eliminates batch effects during cross-study integrationExtensible to scRNA-seq, scATAC-seq, and spatial transcriptomics datasetsIntegrates with scvi-tools and Scanpy ecosystems
scArches uses transfer learning and architectural surgery to seamlessly map new single-cell query datasets onto massive reference atlases (like the Human Cell Atlas) without retraining or sharing raw data.
Information checked against an official source; not a hands-on test. Source · Last reviewed: 20/09/2026, 11:09:40
Key Features
Architectural surgery updating reference neural networks with query-specific parameters
Preserves privacy and eliminates batch effects during cross-study integration
Extensible to scRNA-seq, scATAC-seq, and spatial transcriptomics datasets
Integrates with scvi-tools and Scanpy ecosystems
Academic Context & Research Evidence
Biological & Workflow Fit
Biological Application
Cell-type annotation, patient disease state mapping, and cell atlas expansion
Use this citation format when referencing scArches in scientific publications and benchmark papers.
@software{scarches_2026,
title = {{scArches}},
author = {{Helmholtz Munich (Theis Lab)}},
year = {2026},
url = {https://github.com/theislab/scarches},
note = {Indexed on aibioatlas - AI for Biology and Drug Discovery}
}
Peer-Reviewed Literature & Preprints
Live scientific citations streamed from Europe PMC and PubMed for scArches.
⏳ Fetching real-time literature from Europe PMC & PubMed...
Technical / Product Information
Missing values mean the catalog has no recorded information. They do not mean a feature is absent.