ZipStrain

Fast strain-level metagenomics in three commands: map reads to BAMs, profile them into per-position nucleotide-count tables, and compare samples by ANI to tell whether they share a strain.

Documentation:

• Docs: https://OlmLab.github.io/ZipStrain/
• Repository: https://github.com/OlmLab/ZipStrain

Install

Conda is the easiest path — it brings in samtools and the map aligners too:

conda create -n zipstrain -c conda-forge -c bioconda \
  python=3.12 zipstrain bowtie2 samtools sylph

Or with pip (into a fresh virtual environment):

pip install zipstrain

Matrix-store comparison dependencies (compare --method matrix):

pip install "zipstrain[matrix]"

Notes:

• With pip, install samtools separately (profiling needs it); zipstrain map additionally needs bowtie2 and sylph (and prodigal for --predict-genes).

• On Apple Silicon, use a native osx-arm64 Conda so dependencies install natively; the standard torch wheel uses the MPS backend.

• Linux CUDA installs should replace Torch with the matching CUDA wheel from PyTorch:

  pip install "zipstrain[matrix]"
  pip install --upgrade torch --index-url https://download.pytorch.org/whl/cu124
  

See the installation guide for full details.

Verify

zipstrain --version
zipstrain test

Cite

If you use ZipStrain in your research, please cite the preprint:

Ghadermazi P, Emerson JB, Olm MR. 2026. ZipStrain Enables Rapid and Precise Strain-Resolved Metagenomics. bioRxiv. DOI: 10.64898/2026.05.20.726564

GitHub citation metadata is provided in CITATION.cff:

• https://github.com/OlmLab/ZipStrain/blob/main/CITATION.cff

License

ZipStrain is distributed under the MIT License.