<tool id="sampler" name="Read sampling" version="1.0.1.3">
  <description> Tool for randomly sampling subsets of reads from large datasets</description>
  <requirements>
    <requirement type="package">seqkit</requirement>
    <requirement type="package" version="3">python</requirement>
  </requirements>
  <stdio>
    <exit_code range="1:" level="fatal" description="Error" />
  </stdio>
  <required_files>
    <include type="literal" path="deinterlacer.py" />
    <include type="literal" path="fasta_interlacer.py" />
  </required_files>
  <command>
    <![CDATA[
    #if str($paired)=="true"
      python ${__tool_directory__}/deinterlacer.py $input Afile Bfile
      &&
      NUMBER=\$(($(number) / 2))
      &&
      seqkit sample -2 --number \$NUMBER --rand-seed $seed -o Asample -w 0 Afile < /dev/null
      &&
      seqkit sample -2 --number \$NUMBER --rand-seed $seed -o Bsample -w 0 Bfile < /dev/null
      &&
      python ${__tool_directory__}/fasta_interlacer.py -a Asample -b Bsample -p $output -x tmpfile
    #else
      seqkit sample -2 --number $number --rand-seed $seed -o $output -w 0 $input < /dev/null
    #end if
    ]]>
    </command>

  <inputs>
    <param format="fasta" type="data" name="input" label="Read file (FASTA)" />
    <param name="paired" type="boolean" truevalue="true" falsevalue="false" checked="True" label="Paired-end reads" help="If paired-end reads are sampled, left and right-hand reads must be interlaced and all pairs must be complete."/>
    <param name="number" type="integer" size="7" value="500000" min="1" label="Number of reads"/>
    <param name="seed" type="integer" size="10" value="10" min="0" label="Random number generator seed " />
        
  </inputs>


  <outputs>
 	  <data format="fasta" name="output" label="Random selection from dataset ${input.hid}, sample size ${number})" />
  </outputs>

  <help>
    **What it does**
    
    This tools randomly samples the specified number of reads from larger datasets.
    Using the same random number generator seed with the same dataset results in sampling the same set of reads, while 
    using different seeds generates different subsets of reads. 

    
  </help>

  
</tool>