<tool id="presto_collapseseq" name="pRESTO CollapseSeq" version="@PRESTO_VERSION@">
    <description>Remove/collapse duplicate sequences</description>
    
    <macros>
        <import>presto_macros.xml</import>
    </macros>
    
    <expand macro="requirements"/>
    
    <version_command>CollapseSeq.py --version</version_command>
    <command detect_errors="exit_code"><![CDATA[
        ln -s '$fastq_in' in.fastq &&
        CollapseSeq.py
          -s in.fastq
          -n '$n'
          --uf $uf
          --cf $cf
          --act $act
          $inner
          $keepmiss
          
          #if $representative.select_by == "by_min"
            --minf '$minf'
          #end if
          #if $representative.select_by == "by_max"
            --maxf '$maxf'
          #end if

          --outname=tmp 
          #if $capture_log
            --log '$log_out'
          #end if
    ]]></command>

    <inputs>
        <param argument="-s" name="fastq_in" type="data" format="fastq"
            label="Input FASTQ file" help="FASTQ file of sequences."/>
        <param argument="-n" type="integer" value="0" label="Max Missing" 
            help="Maximum number of missing nucleotides to consider for collapsing sequences. A sequence will be considered undetermined if it contains too many missing nucleotides."/>
        <param argument="--uf" type="text" value="" label="Unique Field(s)" help="Space-separated list of fields to match reads by."/>
        <param argument="--cf" type="text" value="" label="Copy Field(s)" help="Space-separated list of fields to copy into the target sequence headers.">
            <expand macro="text-regex-validator"/>
        </param>
        <param argument="--act" type="text" value="" label="Action(s)" help="Space-separated list of actions to take on Copy Field(s).">
            <expand macro="text-regex-validator"/>
        </param>
        <param argument="--inner" type="boolean" value="false" truevalue="--inner" falsevalue="" label="Inner" help="If true, trim off any missing/undetermined bases from the ends of reads after collapsing."/>
        <param argument="--keepmiss" type="boolean" value="false" truevalue="--keepmiss" falsevalue="" label="Keep Missing" 
            help="If specified, sequences with more missing characters than the threshold set by the -n parameter will be written to the unique sequence output file with a DUPCOUNT=1 annotation. If not specified, such sequences will be written to a separate file."/>
        <conditional name="representative">
            <param name="select_by" type="select">
                <option value="first">First Sequence</option>
                <option value="by_min">By field minimum value</option>
                <option value="by_max">By field maximum value</option>
            </param>
            <when value="first"></when>
            <when value="by_min">
                <param argument="--minf" type="text" value="" label="Field Name" help="Specifies the field whose minimum value determines the retained sequence."/>
            </when>
            <when value="by_max">
                <param argument="--maxf" type="text" value="" label="Field Name" help="Specifies the field whose maximum value determines the retained sequence."/>
            </when>
        </conditional>
        <expand macro="presto-log-param"/>
    </inputs>

    <outputs>
        <data name="fastq_out" format="fastq" from_work_dir="tmp_collapse-unique.fastq"/>
        <expand macro="presto-log-output"/>
    </outputs>
    
    <tests>
        <test>
            <param name="fastq_in" value="presto_collapseseq_test_in.fastq"/>
            <param name="n" value="0"/>
            <param name="uf" value="PRCONS CREGION"/>
            <param name="cf" value="CONSCOUNT"/>
            <param name="act" value="sum"/>
            <param name="inner" value="true"/>
            <param name="keepmiss" value="true"/>
            <output name="fastq_out" file="presto_collapseseq_test_output.fastq" sort="true"/>
        </test>
    </tests>
    
    <help><![CDATA[
Removes/collapses duplicate sequences from FASTQ files.

See the `pRESTO online help <@PRESTO_URL_BASE@/CollapseSeq.html>`_ for more information.

@HELP_NOTE@
    ]]></help>
    <expand macro="citations" />
</tool>