<?xml version="1.0"?>
<tool id="seqtk_sample" name="seqtk_sample" version="@TOOL_VERSION@.1">
    <description>random subsample of fasta or fastq sequences</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="requirements"/>
    <expand macro="stdio"/>
    <command><![CDATA[
seqtk sample
-s $s
'$in_file'
$subsample_size
@CONDITIONAL_GZIP_OUT@
    ]]></command>
    <inputs>
        <expand macro="in_faq" />
        <param argument="-s" type="integer" value="4" label="RNG seed" help="The seed used for the random number generator. Manually specifying a number here is useful for reproducing the same subsampling over time" />
        <param name="subsample_size" type="float" value="100" label="Subsample (decimal fraction or number)" />
    </inputs>
    <outputs>
        <data name="default" format_source="in_file" label="Subsample of reads from $in_file.name"/>
    </outputs>
    <tests>
        <test>
            <param name="in_file" value="seqtk_sample.fa"/>
            <param name="subsample_size" value="4"/>
            <param name="s" value="4"/>
            <output name="default" file="seqtk_sample.out" ftype="fasta"/>
        </test>
        <test>
            <param name="in_file" value="seqtk_sample.fa.gz" ftype="fasta.gz"/>
            <param name="subsample_size" value="4"/>
            <param name="s" value="4"/>
            <output name="default" file="seqtk_sample.out.gz" ftype="fasta.gz"/>
        </test>
    </tests>
    <help><![CDATA[
**What it does**

Takes a random subsample of FASTA or FASTQ sequences. The RNG is seedable to allow for reproducible results, and defaults to `4 <http://xkcd.com/221/>`__.

The subsample size can be a decimal fraction <=1, where 1 implies 100% of the reads should be used. If a number >1 is provided, that many reads will be taken from the dataset.

@ATTRIBUTION@
    ]]></help>
    <expand macro="citation" />
</tool>