<?xml version="1.0"?>
<tool id="seqtk_cutN" name="seqtk_cutN" version="@TOOL_VERSION@.1">
    <description>cut sequence at long N</description>
    <expand macro="bio_tools"/>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="requirements"/>
    <expand macro="stdio"/>
    <command><![CDATA[
seqtk cutN
-n $n
-p $p
$g
'$in_file'
@CONDITIONAL_GZIP_OUT@
    ]]></command>
    <inputs>
        <expand macro="in_faq"/>
        <param argument="-n" type="integer" value="1000" label="Minimum size of N tract" />
        <param argument="-p" type="integer" value="10" label="Penalty for a non-N" />
        <param argument="-g" type="boolean" truevalue="-g" falsevalue="" checked="false" label="Print gaps only, no sequence" />
    </inputs>
    <outputs>
        <data name="default" format_source="in_file" label="$in_file.name split on N runs longer than $n"/>
    </outputs>
    <tests>
        <test>
            <param name="in_file" value="seqtk_cutn.fa"/>
            <param name="n" value="1"/>
            <output name="default" file="seqtk_cutn.out" ftype="fasta"/>
        </test>
        <test>
            <param name="in_file" value="seqtk_cutn.fa.gz" ftype="fasta.gz"/>
            <param name="n" value="1"/>
            <output name="default" file="seqtk_cutn.out.gz" ftype="fasta.gz"/>
        </test>
    </tests>
    <help><![CDATA[
**What it does**

Splits long sequences with runs of Ns

::

    >test
    AACTGATCGATCGATCGNNNNNNNNNNNACATG

This will be split into the component sequences without the ambiguity.

::

    >test:1-17
    AACTGATCGATCGATCG
    >test:29-33
    ACATG

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