Skip to content

Home

Jump to bottom
github-actions[bot] edited this page Oct 22, 2026 · 14 revisions

mashID logo

mashID

Species identification from genome assemblies (fasta) or raw sequencing reads (fastq) with Mash.

mashID runs mash screen for every sample against a sketch database, keeps the best hits, names the organism from the database metadata, and flags results that deserve a second look: possible mixtures, ambiguous calls, low coverage. It works with Illumina (paired-end or single-end), Ion Torrent, Nanopore and PacBio reads, and with assemblies; inputs may be gzipped.

Pages

In one minute

conda create -n mashID -c conda-forge -c bioconda mashid
conda activate mashID
mashID_download_db mycobacteriaceae
mashID -i /path/to/reads -o results

How it works

mashID workflow

  1. Input files are grouped into samples by name; R1/R2 pairs and multi-lane files belong to one sample.
  2. Reads and bases are counted (pure Python, optionally accelerated by python-isal).
  3. mash screen -w estimates, for every reference in the database, how much of its sketch is contained in the sample. Winner-take-all credits shared hashes to the best reference.
  4. Hits to references shorter than 100 kb are ignored (partial records), the rest are ranked, and the organism name, TaxID and reference length are read from the database's metadata sidecar.
  5. Notes are added from the hit list, and the summary (TSV, JSON, MultiQC), per-sample tables and a run-provenance file are written.

These pages are generated from docs/wiki in the repository. Edit them there; a workflow publishes changes to this wiki.

mashID

Releases · Changelog · Issues

Clone this wiki locally