Arcane¶
Alignment-free single-cell RNA-seq gene expression estimation.
Arcane is a lightweight tool that turns scRNAseq data with barcode and UMI information, such as 10X genomics or DropSeq data, into a gene-by-cell count matrix, without aligning reads to a reference genome. Instead of alignment, it looks up gapped k-mers in a bucketed Cuckoo hash table that stores, for each k-mer, the (few) genes it can originate from.
A complete run is one command:
arcane express --index myindex --R1 $R1_files --R2 $R2_files \
--out outfolder -c v3 --kneemethod distance
See the Quickstart to go from a prebuilt index to a count matrix.
What it does¶
Arcane processes a sample in three stages, which you can run as one command
(arcane express) or individually:
| Stage | Command | Input | Output |
|---|---|---|---|
| Barcode correction | arcane correct |
R1 (barcode + UMI) | corrected barcodes, cell calls |
| Mapping | arcane map |
R1 + R2 (cDNA sequence) | UMI-gene assignments |
| UMI resolution | arcane count |
the two above | count matrix |
Building an index is a separate, one-off step (arcane filter
then arcane index) — and for human and mouse you can skip it
by downloading a prebuilt index.
Where to go next¶
- Installation — conda environment,
pip install, prebuilt indices. - Quickstart — the shortest path to a count matrix.
- Building an index — custom references, masks, table sizing.
- Running a sample — chemistries, mapping modes, UMI modes.
- Bulk mode — quantifying bulk RNA seq experiment.
- Shared memory — run many samples without paying for the index each time.
- CLI reference — every subcommand and flag.
- File formats — what each input and output file contains.
Support¶
Please file an issue in the issue tracker.
Recent changes are listed in CHANGELOG.md in the repository.
Citation¶
If you use Arcane, please cite:
Zentgraf, J., Schmitz, J. E., Keller, A., Rahmann, S. Error Correction Algorithms for Efficient Gene Expression Quantification in Single Cell Transcriptomics. bioRxiv (2025). doi:10.1101/2025.11.27.690682
License¶
MIT. (c) 2019-2026 Algorithmic Bioinformatics, Saarland University.