comparison repeatmasker.xml @ 10:bfc70c8cc5ca draft

"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/repeat_masker commit 69abffb867af7c4329eaa513201bd6626ac39cbf"

9:438f65cb1d14

<tool id="repeatmasker_wrapper" name="RepeatMasker" version="4.0.9" profile="17.01">
  <description>screen DNA sequences for interspersed repeats and low complexity regions</description>

  <requirements>
    <requirement type="package" version="4.0.9_p2">repeatmasker</requirement>
  </requirements>

  <command detect_errors="exit_code"><![CDATA[
    RM_PATH=\$(which RepeatMasker) &&
    if [ -z "\$RM_PATH" ] ; then echo "Failed to find RepeatMasker in PATH (\$PATH)" >&2 ; exit 1 ; fi &&
    RM_LIB_PATH=\$(dirname \$RM_PATH)/../share/RepeatMasker/Libraries &&
    mkdir lib &&
    export REPEATMASKER_LIB_DIR=\$(pwd)/lib &&
      for file in \$(ls \$RM_LIB_PATH) ; do  ln -s \$RM_LIB_PATH/\$file lib/\$file ; done &&
    #if $repeat_source.source_type == "repbase":
      cp '${repeat_source.repbase_file}' 'lib/${repeat_source.repbase_file_name}' &&
    #end if
    ln -s '${input_fasta}' rm_input.fasta &&
    RepeatMasker -dir \$(pwd)
    #if $repeat_source.source_type == "library":
      -lib '${repeat_source.repeat_lib}'
      -cutoff '${repeat_source.cutoff}'
    #else if $repeat_source.source_type == "repbase":
      #if $repeat_source.species_source.species_from_list == 'yes':
        $repeat_source.species_source.species_list
      #else
        -species '${repeat_source.species_source.species_name}'
      #end if
    #end if
    -parallel \${GALAXY_SLOTS:-1}
    ${gff}
    ${excln}
    ${advanced.is_only}

  </command>

  <inputs>
    <param name="input_fasta" type="data" format="fasta" label="Genomic DNA" />
    <conditional name="repeat_source">
      <param label="Repeat library source" name="source_type" type="select">
        <option selected="true" value="repbase">RepBase</option>
        <option value="library">Custom library of repeats</option>
      </param>
      <when value="repbase">
        <param name="repbase_file" type="data" format="embl" label="RepBase (RMRBSeqs.embl) file" />
        <param name="repbase_file_name" type="hidden" value="RMRBSeqs.embl"/> <!-- This is an ugly hack to allow testing with a fake repbase -->
        <conditional name="species_source">
          <param label="Select species name from a list?" name="species_from_list" type="select">
            <option value="yes" selected="true">Yes</option>
            <option value="no">No</option>
          </param>
          <when value="yes">
            <param name="species_list" type="select" label="Species">
              <option value="-species anopheles" selected="true">anopheles</option>
              <option value="-species arabidopsis">arabidopsis</option>
              <option value="-species artiodactyl">artiodactyl</option>
              <option value="-species aspergillus">aspergillus</option>
              <option value="-species carnivore">carnivore</option>
              <option value="-species cat">cat</option>
              <option value="-species chicken">chicken</option>
              <option value="-species 'ciona intestinalis'">ciona intestinalis</option>
              <option value="-species 'ciona savignyi'">ciona savignyi</option>
              <option value="-species cow">cow</option>
              <option value="-species danio">danio</option>
              <option value="-species diatoaea">diatomea</option>
              <option value="-species dog">dog</option>
              <option value="-species drosophila">drosophila</option>
              <option value="-species elegans">elegans</option>
              <option value="-species fugu">fugu</option>
              <option value="-species fungi" selected="true">fungi</option>
              <option value="-species human">human</option>
              <option value="-species maize">maize</option>
              <option value="-species mammal">mammal</option>
              <option value="-species mouse">mouse</option>
              <option value="-species pig">pig</option>
              <option value="-species rat">rat</option>
              <option value="-species rice">rice</option>
              <option value="-species rodentia">rodentia</option>
              <option value="-species ruminantia">ruminantia</option>
              <option value="-species wheat">wheat</option>
            </param>
          </when>
          <when value="no">
            <param name="species_name" type="text" value="homo sapiens" label="Repeat source species" help="Source species (or clade name) used to select repeats from RepBase" />
          </when>
        </conditional>
      </when>
      <when value="library">
        <param name="repeat_lib" type="data" format="fasta" label="Custom library of repeats" />
        <param name="cutoff" type="integer" argument="-cutoff" value="225" label="Cutoff score for masking repeats" />
      </when>

    </test>
    <test expect_num_outputs="7">
      <param name="input_fasta" value="small.fasta" ftype="fasta" />
      <param name="source_type" value="library" />
      <param name="gff" value="-gff" />
      <!-- <param name="show" value="yes" /> -->
      <param name="keep_alignments" value="-ali" />
      <param name="poly" value="-poly" />
      <param name="repeat_lib" value="repeats.fasta" ftype="fasta" />
      <output name="output_masked_genome" file="small.fasta.masked" />
      <output name="output_table" file="small.fasta.stats" lines_diff="6" />

      <output name="output_polymorphic" file="small.fasta.poly" />
      <output name="output_gff" file="small.fasta.gff" lines_diff="4" />
    </test>
    <test expect_num_outputs="4">
      <param name="input_fasta" value="small.fasta" ftype="fasta" />
      <param name="source_type" value="repbase" />
      <param name="repbase_file" value="fake_repbase.embl" />
      <param name="repbase_file_name" value="fake.embl" />
      <param name="species_list" value="anopheles" />
      <output name="output_masked_genome" file="small.fasta.masked" />
      <output name="output_table" file="small_repbase.fasta.stats" lines_diff="2" />
      <output name="output_repeat_catalog" file="small.fasta.cat" lines_diff="2" />
      <output name="output_log" file="small_repbase.fasta.log" lines_diff="2"/>
    </test>
  </tests>
  <help><![CDATA[
RepeatMasker is a program that screens DNA for interspersed repeats and low
complexity DNA sequences. The database of repeats to screen for can be

10:bfc70c8cc5ca

<tool id="repeatmasker_wrapper" name="RepeatMasker" version="4.1.1" profile="17.01">
  <description>screen DNA sequences for interspersed repeats and low complexity regions</description>

  <requirements>
    <requirement type="package" version="4.1.1">repeatmasker</requirement>
  </requirements>

  <command detect_errors="exit_code"><![CDATA[
    RM_PATH=\$(which RepeatMasker) &&
    if [ -z "\$RM_PATH" ] ; then echo "Failed to find RepeatMasker in PATH (\$PATH)" >&2 ; exit 1 ; fi &&

    RM_LIB_PATH=\$(dirname \$RM_PATH)/../share/RepeatMasker/Libraries &&
    #if $repeat_source.source_type == "dfam_up":
      mkdir lib/ &&
      ln -s '${repeat_source.dfam_lib}' lib/RepeatMaskerLib.h5 &&
      RM_LIB_PATH=\$(pwd)/lib &&
    #end if

    ln -s '${input_fasta}' rm_input.fasta &&

    RepeatMasker -dir \$(pwd)
    -libdir \$RM_LIB_PATH
    #if $repeat_source.source_type == "library":
      -lib '${repeat_source.repeat_lib}'
      -cutoff '${repeat_source.cutoff}'
    #else if $repeat_source.source_type == "dfam":
      #if $repeat_source.species_source.species_from_list == 'yes':
        -species $repeat_source.species_source.species_list
      #else
        -species '${repeat_source.species_source.species_name}'
      #end if
    #else if $repeat_source.source_type == "dfam_up":
        -species '${repeat_source.species_name}'
    #end if
    -parallel \${GALAXY_SLOTS:-1}
    ${gff}
    ${excln}
    ${advanced.is_only}

  </command>

  <inputs>
    <param name="input_fasta" type="data" format="fasta" label="Genomic DNA" />
    <conditional name="repeat_source">
      <param label="Repeat library source" name="source_type" type="select" help="To use RepBase, choose 'Custom library of repeats' and select a fasta version of this non-free database.">
        <option selected="true" value="dfam">DFam (curated only, bundled with RepeatMasker)</option>
        <option value="dfam_up">DFam (full/specific version)</option>
        <option value="library">Custom library of repeats</option>
      </param>
      <when value="dfam">
        <conditional name="species_source">
          <param label="Select species name from a list?" name="species_from_list" type="select">
            <option value="yes" selected="true">Yes</option>
            <option value="no">No</option>
          </param>
          <when value="yes">
            <param name="species_list" type="select" label="Species">
              <option value="vertebrate">Vertebrate (other than below)</option>
              <option value="mammal">Mammal (other than below)</option>
              <option value="human" selected="true">Human</option>
              <option value="rodent">Rodent</option>
              <option value="mouse">Mouse</option>
              <option value="rat">Rat</option>
              <option value="danio">Danio (zebra fish)</option>
              <option value="drosophila">Fruit fly (Drosophila melanogaster)</option>
              <option value="elegans">Caenorhabditis elegans (nematode)</option>
            </param>
          </when>
          <when value="no">
            <param name="species_name" type="text" value="human" label="Repeat source species" help="Source species (or clade name) used to select repeats from DFam" />
          </when>
        </conditional>
      </when>
      <when value="dfam_up">
          <param name="dfam_lib" type="data" format="h5" label="DFam library" help="The full DFam library can be downloaded from https://www.dfam.org/releases/current/families/Dfam.h5.gz" />
          <param name="species_name" type="text" value="human" label="Repeat source species" help="Source species (or clade name) used to select repeats from DFam" />
      </when>
      <when value="library">
        <param name="repeat_lib" type="data" format="fasta" label="Custom library of repeats" />
        <param name="cutoff" type="integer" argument="-cutoff" value="225" label="Cutoff score for masking repeats" />
      </when>

    </test>
    <test expect_num_outputs="7">
      <param name="input_fasta" value="small.fasta" ftype="fasta" />
      <param name="source_type" value="library" />
      <param name="gff" value="-gff" />
      <param name="keep_alignments" value="-ali" />
      <param name="poly" value="-poly" />
      <param name="repeat_lib" value="repeats.fasta" ftype="fasta" />
      <output name="output_masked_genome" file="small.fasta.masked" />
      <output name="output_table" file="small.fasta.stats" lines_diff="6" />

      <output name="output_polymorphic" file="small.fasta.poly" />
      <output name="output_gff" file="small.fasta.gff" lines_diff="4" />
    </test>
    <test expect_num_outputs="4">
      <param name="input_fasta" value="small.fasta" ftype="fasta" />
      <param name="source_type" value="dfam" />
      <param name="species_list" value="human" />
      <output name="output_masked_genome" file="small_dfam.fasta.masked" />
      <output name="output_table" file="small_dfam.fasta.stats" lines_diff="2" />
      <output name="output_repeat_catalog" file="small_dfam.fasta.cat" lines_diff="2" />
      <output name="output_log" file="small_dfam.fasta.log" lines_diff="2"/>
    </test>
    <test expect_num_outputs="4">
      <param name="input_fasta" value="small.fasta" ftype="fasta" />
      <param name="source_type" value="dfam_up" />
      <param name="dfam_lib" value="Dfam_partial_test.h5" ftype="h5" />
      <param name="species_name" value="rodent" />
      <output name="output_masked_genome" file="small_dfam_up.fasta.masked" />
      <output name="output_table" file="small_dfam_up.fasta.stats" lines_diff="2" />
      <output name="output_repeat_catalog" file="small_dfam_up.fasta.cat" lines_diff="2" />
      <output name="output_log" file="small_dfam_up.fasta.log" lines_diff="2"/>
    </test>
  </tests>
  <help><![CDATA[
RepeatMasker is a program that screens DNA for interspersed repeats and low
complexity DNA sequences. The database of repeats to screen for can be