name: "mash_screen"
description: Screens query sequences against large sequence databases
keywords:
  - screen
  - containment
  - contamination
  - taxonomic assignment
tools:
  - "mash":
      description: Fast sequence distance estimator that uses MinHash
      homepage: https://github.com/marbl/Mash
      documentation: https://mash.readthedocs.io/en/latest/sketches.html
      tool_dev_url: https://github.com/marbl/Mash
      doi: "10.1186/s13059-016-0997-x"
      licence: ["https://github.com/marbl/Mash/blob/master/LICENSE.txt"]

input:
  - meta:
      type: map
      description: |
        Groovy Map containing sample information
        e.g. [ id:'test', single_end:false ]
  - query:
      type: file
      description: Query sequences
      pattern: "*.fastq.gz"
  - sequence_sketch:
      type: file
      description: Sequence files to match against
      pattern: "*.msh"

output:
  - meta:
      type: map
      description: |
        Groovy Map containing sample information
        e.g. [ id:'test', single_end:false ]
  - versions:
      type: file
      description: File containing software versions
      pattern: "versions.yml"
  - screen:
      type: file
      description: List of sequences from fastx_db similar to query sequences
      pattern: "*.screen"

authors:
  - "@mahesh-panchal"