name: bakta
description: Annotation of bacterial genomes (isolates, MAGs) and plasmids
keywords:
  - annotation
  - fasta
  - bacteria
tools:
  - bakta:
      description: Rapid & standardized annotation of bacterial genomes, MAGs & plasmids.
      homepage: https://github.com/oschwengers/bakta
      documentation: https://github.com/oschwengers/bakta
      tool_dev_url: https://github.com/oschwengers/bakta
      doi: "10.1099/mgen.0.000685"
      licence: ["GPL v3"]

input:
  - meta:
      type: map
      description: |
        Groovy Map containing sample information
        e.g. [ id:'test', single_end:false ]
  - fasta:
      type: file
      description: |
        FASTA file to be annotated. Has to contain at least a non-empty string dummy value.
  - db:
      type: file
      description: |
        Path to the Bakta database
  - proteins:
      type: file
      description: FASTA/GenBank file of trusted proteins to first annotate from (optional)
  - prodigal_tf:
      type: file
      description: Training file to use for Prodigal (optional)

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"
  - txt:
      type: file
      description: genome statistics and annotation summary
      pattern: "*.txt"
  - tsv:
      type: file
      description: annotations as simple human readble tab separated values
      pattern: "*.tsv"
  - gff:
      type: file
      description: annotations & sequences in GFF3 format
      pattern: "*.gff3"
  - gbff:
      type: file
      description: annotations & sequences in (multi) GenBank format
      pattern: "*.gbff"
  - embl:
      type: file
      description: annotations & sequences in (multi) EMBL format
      pattern: "*.embl"
  - fna:
      type: file
      description: replicon/contig DNA sequences as FASTA
      pattern: "*.fna"
  - faa:
      type: file
      description: CDS/sORF amino acid sequences as FASTA
      pattern: "*.faa"
  - ffn:
      type: file
      description: feature nucleotide sequences as FASTA
      pattern: "*.ffn"
  - hypotheticals_tsv:
      type: file
      description: additional information on hypothetical protein CDS as simple human readble tab separated values
      pattern: "*.hypotheticals.tsv"
  - hypotheticals_faa:
      type: file
      description: hypothetical protein CDS amino acid sequences as FASTA
      pattern: "*.hypotheticals.faa"

authors:
  - "@rpetit3"
  - "@oschwengers"