<tool id="art_454" name="ART 454" version="2014.11.03.0">
  <description>simulates pyrosequencing data</description>
  <macros>
      <import>macros.xml</import>
  </macros>
  <expand macro="requirements"/>
  <expand macro="stdio" />
  <command><![CDATA[
art_454 $t

$aln
$sam

#if $rndSeed and $rndSeed > -1:
-r $rndSeed
#end if

#if $c
-c $c
#end if

#if $generate.amplicon.use_amplicon == "amplicon_true":
    #if $generate.choice == "single_end":
        -A
    #else:
        -B
    #end if
#end if

$input_seq_file
output

#if $generate.choice == "single_end":
    $fold_coverage
#else:
    $fold_coverage
    $generate.fragment_size
    $generate.fragment_sd
#end if

#if $generate.amplicon.use_amplicon == "amplicon_true":
    #if $generate.choice == "single_end":
        $generate.amplicon.reads_per_amplicon
    #else:
        $generate.amplicon.read_pairs_per_amplicon
    #end if
#end if
;
]]></command>
  <inputs>
    <param label="DNA/RNA reference sequence" format="fasta" name="input_seq_file" type="data"/>
    <param label="the fold of read coverage over the reference sequences" name="fold_coverage" type="integer" value="20"/>

    <conditional name="generate">
        <param name="choice" type="select" label="Type of data to generate">
            <option value="single_end">Single-End</option>
            <option value="paired_end">Paired-End</option>
        </param>
        <when value="single_end">
            <expand macro="amplicon" />
        </when>
        <when value="paired_end">
            <expand macro="frag_len_sd" />
            <expand macro="amplicon_pair" />
        </when>
    </conditional>


    <expand macro="sam" />
    <expand macro="aln" />

    <param type="boolean" label="indicate to simulate reads from the built-in GS FLX Titanium profile (-t)" name="t" truevalue="-t" falsevalue="" optional="true"  />
    <param label="specify the number of flow cycles by the sequencer [100 for GS-FLX, 200 for GS-FLX Titanium] (-c)" name="c" type="integer" value="100" optional="true" />

    <expand macro="rndSeed" />
  </inputs>
  <outputs>
        <!-- Single End -->
        <data format="fastq" name="output_fq1_single" from_work_dir="output.fq" label="Simulated of 454 sequencing of $input_seq_file.name">
            <filter>generate['choice'] == "single_end"</filter>
        </data>

        <!-- Paired End -->
        <data format="fastq" name="output_fq1_paired" from_work_dir="output1.fq" label="Simulated of 454 sequencing of $input_seq_file.name (Forward)">
            <filter>generate['choice'] != "single_end"</filter>
        </data>
        <data format="fastq" name="output_fq2_paired" from_work_dir="output2.fq" label="Simulated of 454 sequencing of $input_seq_file.name (Reverse)">
            <filter>generate['choice'] != "single_end"</filter>
        </data>
        <data format="sam" name="output_sam" from_work_dir="output.sam" label="Mapping of Simulated 454 data to $input_seq_file.name">
            <filter>sam</filter>
        </data>

        <!-- Single End -->
        <data format="aln" name="output_aln1_single" from_work_dir="output.aln" label="Alignment of Simulated 454 data to $input_seq_file.name">
            <filter>aln and generate['choice'] == "single_end"</filter>
        </data>
        <!-- Paired End -->
        <data format="aln" name="output_aln1_paired" from_work_dir="output1.aln" label="Alignment of Simulated 454 data to $input_seq_file.name">
            <filter>aln and generate['choice'] != "single_end"</filter>
        </data>
        <data format="aln" name="output_aln2_paired" from_work_dir="output2.aln" label="Alignment of Simulated 454 data to $input_seq_file.name">
            <filter>generate['choice'] != "single_end" and generate['amplicon']['use_amplicon'] == "amplicon_true"</filter>
        </data>
  </outputs>
  <tests>
      <!--  Single End tests -->
      <test>
          <param name="rndSeed" value="42" />
          <param name="input_seq_file" value="input.fa" />
          <param name="fold_coverage" value="20" />
          <param name="choice" value="single_end" />
          <output name="output_fq1_single" file="art.454.01.fq" compare="sim_size" delta="5000" />
      </test>
      <test>
          <param name="rndSeed" value="42" />
          <param name="input_seq_file" value="input.fa" />
          <param name="fold_coverage" value="20" />
          <param name="choice" value="single_end" />
          <param name="sam" value="True" />
          <output name="output_fq1_single" file="art.454.01.fq" compare="sim_size" delta="5000" />
          <output name="output_sam" file="art.454.01.sam" compare="sim_size" delta="5000"/>
      </test>
      <!-- Paired End tests -->
      <test>
          <param name="rndSeed" value="42" />
          <param name="input_seq_file" value="input.fa" />
          <param name="fold_coverage" value="20" />
          <param name="choice" value="paired_end" />
          <param name="fragment_size" value="105" />
          <param name="fragment_sd" value="5" />
          <param name="sam" value="True" />
          <output name="output_fq1_paired" file="art.454.021.fq" compare="sim_size" delta="5000" />
          <output name="output_fq2_paired" file="art.454.022.fq" compare="sim_size" delta="5000" />
          <output name="output_sam" file="art.454.02.sam" compare="sim_size" delta="5000"/>
      </test>
  </tests>
  <help><![CDATA[
Art 454 Pyrosequencing Simulator
================================

ART_454 is a simulation program to generate sequence read data of Roche 454
Pyrosequencing sequencers. ART generates reads according to the empirical read
quality profile and the calibrated error profile of uncall/overcall
homopolymers from real 454 read data. ART has been using for testing or
benchmarking a variety of method or tools for next-generation sequencing data
analysis, including read alignment, de novo assembly, detection of SNP, CNV, or
other structure variation.

art_454 can generate both single-end and paired-end of 454 sequencing platform.
Besides for regular genome DNA and cDNA sequencing simulation, art_454 also
supports amplicon sequencing. The reference sequences can be either DNA or RNA.
      ]]></help>
  <expand macro="citation" />
</tool>