<tool id="ectyper" name="ectyper" version="0.9.0">
  <description>ectyper is a standalone serotyping module for Escherichia coli. It supports fasta and fastq file formats.</description>
  <requirements>
     <requirement type="package" version="0.9.0">ectyper</requirement>
  </requirements>
  <command detect_errors="exit_code">
  <![CDATA[
  #set $genomes = ''
  #if hasattr($input, '__iter__')
    #for $i in $input
        ln -s "${i}" "${i.element_identifier}" &&
        #if len($genomes) > 0
          #set $genomes = $genomes + ',' + str($i.element_identifier)
        #else
          #set $genomes = str($i.element_identifier)
        #end if      
    #end for
  #else
    ln -s "${input}" "${input.element_identifier}" &&
    #set $genomes = $input.element_identifier
  #end if

  #if $mash_input
    ln -s "${mash_input}" mash_sketch.msh &&
  #end if

  ectyper  --cores \${GALAXY_SLOTS:-4} 
  --input "${genomes}" 
  --percentIdentity '$adv_param.min_percentIdentity'
  --percentLength '$adv_param.percentLength'
  #if $adv_param.verifyEcoli
    --verify
  #end if
  #if $mash_input
    --refseq mash_sketch.msh
  #end if   
  #if $adv_param.alleleSequence
    --sequence
  #end if
  --output '.'
  ]]>
  </command>
  <inputs>
    <param name="input" type="data"  format="fastq,fasta" label="Genome(s) input(s)" help="FASTA or FASTQ file(s)"/>
    <param name="mash_input" type="data" optional="true" format="binary" label="Mash genome sketches (Optional)" help="Optionally provide custom MASH genome sketch to help with species identification (otherwise default RefSeq sketch is used)"/>
    <section name="adv_param" title="Advanced parameters" expanded="False">
      <param name="min_percentIdentity" type="integer" value="90" min="1" max="100"/>
      <param name="percentLength" type="integer" value="10" min="1" max="100"/>
      <param name="verifyEcoli" type="boolean" checked="true" label="Enable E. coli species verification"/>
      <param name="alleleSequence" type="boolean" checked="false" label="Print the allele sequences as the final columns of the output?"/> 
      <param name="logging" type="boolean" checked="false"  label="Include log file in the run outputs?" />
    </section>  
  </inputs>
  <outputs>
    <data name="output_result" format="tabular" from_work_dir="output.tsv" label="${tool.name} serotype report on ${input.element_identifier}"> </data>
    <data name="output_log" format="txt" from_work_dir="ectyper.log" label="${tool.name} log file on ${input.element_identifier}">
        <filter>adv_param['logging']==True</filter>
    </data>   
  </outputs>
  <tests>
    <test>
      <param name="input" value="Escherichia2.fastq"/>
      <assert_stderr>
            <has_text text="O22"/> 
            <has_text text="H8"/> 
      </assert_stderr>
      <output name="output_result" ftype="tabular" >
          <assert_contents>
              <has_text_matching expression="O22"/>
         </assert_contents>
      </output>
    </test>
  </tests>

  <help>
**Syntax**


This tool identifies the serotype of assembled or assembly-free Escherichia coli genome sample based on a set of either *wzm/wzt* or *wzx/wzy* and *fliC/flkA/flmA* alleles corresponding to O and H antigens, respectively.
The non-E.coli genomes and other Escherichia genus species are successfully identified and well handled. The 0.9.0 version improves tool sensitivy when target alleles are truncated or
poorly covered by raw reads.


For more information please visit https://github.com/phac-nml/ecoli_serotyping. 

-----

**Input:**

Accepts a variety of inputs including both single and/or multiple FASTQ and/or FASTA file(s). Inputs might contain pure raw reads, but for more accurate results, draft assemblies are recommended.

The default MASH RefSeq genome sketch is included and updated every 6 months, but one can supply custom sketch file for species identification.

One can download RefSeq genome sketch containing approximately 91,283 genomes from https://gembox.cbcb.umd.edu/mash/refseq.genomes.k21s1000.msh.


**Output:**

Tab-delimited report listing identified O and H antigens together with corresponding highest scoring alleles and normalized BLAST score defined as (%identity x query coverage length) / 10000

-----

**Parameters (Optional):**

  - **Print the allele sequences as the final columns of the output?** Turn ON/OFF addition of the actual O and H antigen allelic sequences in the report
  - **Enable E. coli species verification:** Turn ON/OFF for more rigorous species verification (recommended)
  - **Include log file in the run outputs?:** Turn ON/OFF optional output of the ectyper log file for a more detailed results assessment

  </help>
<citations>
    <citation type="bibtex">
  @misc{githubectyper,
  author = {Laing Chad},
  title = {ECtyper - serotyping module for Escherichia coli},
  publisher = {GitHub},
  journal = {GitHub repository},
  url = {https://github.com/phac-nml/ecoli_serotyping}
    }</citation>
</citations>
</tool>