vecmap 1.0.0

Ultrafast exact sequence matching in pure Python using NumPy vectorization

VecMap

Ultrafast exact sequence matching using NumPy vectorization. Designed for CRISPR screens and barcode mapping where exact matching is biologically required.

Paper: bioRxiv preprint

Installation

pip install vecmap

Quick Start

# Command line
vecmap -r reference.fa -q reads.fq -o alignments.txt

# Python API
from vecmap import vecmap
alignments = vecmap(reference_seq, [(read_seq, read_id), ...])

Performance

Benchmarks on human transcriptome (42,027 ± 1,856 reads/second in pure Python):

Tool	Reads/sec (mean ± SD)	Language	Notes
VecMap	42,027 ± 1,856	Python	Exact matching only
Minimap2	173,460 ± 5,203	C	General purpose aligner
BWA-MEM	60,306 ± 2,418	C	General purpose aligner

For CRISPR screening specifically:

• VecMap: 18,948 ± 892 reads/sec
• 1.9× faster than MAGeCK
• 3.8× faster than CRISPResso2

Performance measured on Apple M1 Max, 32GB RAM, Python 3.11.5, NumPy 2.0.0

Applications

CRISPR Guide Detection

from vecmap.applications import CRISPRGuideDetector

guides = {
    "KRAS_sg1": "ACGTACGTACGTACGTACGT",
    "TP53_sg1": "GGCCGGCCGGCCGGCCGGCC"
}

detector = CRISPRGuideDetector(guides)
results = detector.detect_guides(reads)
counts = detector.summarize_detection(results)

Barcode Demultiplexing

from vecmap.applications import BarcodeProcessor

processor = BarcodeProcessor(
    barcode_whitelist=whitelist,
    barcode_length=16,
    umi_length=10
)

corrected = processor.correct_barcodes(processor.extract_barcodes(reads))

Reproducibility

To reproduce all benchmarks and figures from the paper:

git clone https://github.com/the-jordan-lab/VecMap.git
cd VecMap
git checkout v1.0.0
pip install -e .
./reproduce.sh

See DATA_AVAILABILITY.md for complete reproduction details.

Documentation

• Examples - Usage examples and tutorials
• Benchmarks - Performance benchmarks and comparisons
• API Reference - Full API documentation

Citation

@article{jordan2025vecmap,
  title={VecMap: Ultrafast Exact Sequence Matching for CRISPR Screens},
  author={Jordan, James M},
  journal={bioRxiv},
  year={2025},
  doi={10.1101/2025.XX.XX.XXXXXX}
}

License

MIT License. See LICENSE for details.