comparison repeatmasker.xml @ 4:04f5c3d7448e draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/repeat_masker commit cbba6947f1380751e1db3fa5b043af630523fd86

3:bdfc22c1c3e3

<tool id="repeatmasker_wrapper" name="RepeatMasker" version="4.0.7" profile="17.01">
  <description>RepeatMasker</description>

  <requirements>
    <requirement type="package" version="4.0.7">repeatmasker</requirement>
  </requirements>

    RM_LIB_PATH=\$(dirname \$(which RepeatMasker))/../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/RMRBSeqs.embl &&
    #end if
    ln -s '${input_fasta}' rm_input.fasta &&
    RepeatMasker -dir \$(pwd)
    #if $repeat_source.source_type == "library":
      -lib '${repeat_source.repeat_lib}'

      #else
        -species '${repeat_source.species_source.species_name}'
      #end if
    #end if
    -parallel \${GALAXY_SLOTS:-1}
    '${gff}'
    '${ignore_n_stretches}'
    '${advanced.is_only}'
    '${advanced.is_clip}'
    '${advanced.no_is}'
    '${advanced.rodspec}'
    '${advanced.primspec}'
    '${advanced.nolow}'
    '${advanced.noint}'
    '${advanced.norna}'
    '${advanced.alu}'
    '${advanced.div}'
    '${advanced.search_speed}'
    '${advanced.frag}'
    '${advanced.maxsize}'
    #if $advanced.gc is not None:
      '${advanced.gc}'
    #end if
    '${advanced.gccalc}'
    '${advanced.nocut}'
    '${advanced.keep_alignments}'
    '${advanced.invert_alignments}'
    '${advanced.xout}'
    '${advanced.xsmall}'
    '${advanced.poly}'
    rm_input.fasta &&
    #if $advanced.is_only != '-is_only':
      mv rm_input.fasta.masked '${output_masked_genome}' &&
      sed -r 's/^ *// ; s/ *$//; s/\+ //; s/ +/\t/g ;  1,2c SW score\t% div.\t% del.\t% ins.\tquery sequence\tpos in  query: begin\tend\t(left)\trepeat\tclass/family\tpos in repeat: begin\tend\t(left)\tID' rm_input.fasta.out >'${output_log}' &&
      mv rm_input.fasta.tbl '${output_table}' &&

        <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" />
        <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>

        <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>
    </conditional>
    <param type="boolean" argument="-gff" truevalue="-gff" falsevalue="" label="Output annotation of repeats in GFF format" checked="false" />
    <param name="ignore_n_stretches" type="boolean" argument="-excln" falsevalue="" label="Ignore stretches of Ns when computing statistics" checked="true" help="Scaffolds are sometimes joined with stretches of 25 or more Ns. This option ignores them when calculating repeat statistics" />
    <section name="advanced" title="Advanced options" expanded="false">
      <param argument="-is_only" type="boolean" truevalue="-is_only" falsevalue="" checked="false" label="Only clip E coli insertion elements" />
      <param argument="-is_clip" type="boolean" truevalue="-is_clip" falsevalue="" checked="false" label="Clip IS elements before analysis" help="Normally RepeatMasker will report on IS element, with this option selected it will clip them before analysis" />
      <param argument="-no_is" type="boolean" truevalue="-no_is" falsevalue="" checked="false" label="Skip bacterial insertion element check" />
      <param argument="-rodspec" type="boolean" truevalue="-rodspec" falsevalue="" checked="false" label="Only check for rodent specific repeats" help="If this option is select a check for rodent specific repeats is done instead of a full RepeatMasker run" />

        <option value="-q">Quick (5-10% less sensitive, 3-4 times speedup)</option>
        <option value="-qq">Rush (10% less sensitive)</option>
        <option value="-s">Slow (0-5% more sensitive, 2.5 times slowdown)</option>
      </param>
      <param type="integer" argument="-frag" value="40000" label="Maximum contiguous sequence searched" help="Maximum length of sequencing that is search without fragmenting" />
      <param type="integer" argument="-maxsize" value="4000000" label="Maximum length for IS or repeat clipped sequences" />
      <param type="integer" argument="-gc" optional="True" label="Select matrices for this GC%" help="Valid values are a percentage or -1 to choose the default" />
      <param type="boolean" argument="-gccalc" truevalue="-gcccalc" falsevalue="" checked="false" label="Calculate GC % for all sequences" help="By default RepeatMasker skips calculating GC % for small sequences" />
      <param type="boolean" argument="-nocut" truevalue="-nocut" falsevalue="" checked="false" label="Skips cutting of repeats" />
      <param name="xout" type="boolean" argument="-x" truevalue="-x" falsevalue="" checked="false" label="Mask with X instead of N characters" />
      <param name="keep_alignments" type="boolean" argument="-ali" truevalue="-ali" falsevalue="" checked="false" label="Output alignments file" />
      <param name="invert_alignments" type="boolean" argument="-inv" truevalue="-inv" falsevalue="" checked="false" label="Invert alignments in alignment file" help="Show alignments in the orientation of the repeat sequence, not the query sequence" />

      <output name="output_log" file="small.fasta.log" />
      <output name="output_alignment" file="small.fasta.align" />
      <output name="output_polymorphic" file="small.fasta.poly" />
      <output name="output_gff" file="small.fasta.gff" lines_diff="4" />
    </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
provided as a FASTA file or downloaded from RepBase_. If the RepBase option is

4:04f5c3d7448e

<tool id="repeatmasker_wrapper" name="RepeatMasker" version="4.0.7+galaxy1" profile="17.01">
  <description>RepeatMasker</description>

  <requirements>
    <requirement type="package" version="4.0.7">repeatmasker</requirement>
  </requirements>

    RM_LIB_PATH=\$(dirname \$(which RepeatMasker))/../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}'

      #else
        -species '${repeat_source.species_source.species_name}'
      #end if
    #end if
    -parallel \${GALAXY_SLOTS:-1}
    ${gff}
    ${excln}
    ${advanced.is_only}
    ${advanced.is_clip}
    ${advanced.no_is}
    ${advanced.rodspec}
    ${advanced.primspec}
    ${advanced.nolow}
    ${advanced.noint}
    ${advanced.norna}
    ${advanced.alu}
    ${advanced.div}
    ${advanced.search_speed}
    -frag ${advanced.frag}
    ## -maxsize ${advanced.maxsize}
    #if str($advanced.gc):
      -gc ${advanced.gc}
    #end if
    ${advanced.gccalc}
    ${advanced.nocut}
    ${advanced.keep_alignments}
    ${advanced.invert_alignments}
    ${advanced.xout}
    ${advanced.xsmall}
    ${advanced.poly}
    rm_input.fasta &&
    #if $advanced.is_only != '-is_only':
      mv rm_input.fasta.masked '${output_masked_genome}' &&
      sed -r 's/^ *// ; s/ *$//; s/\+ //; s/ +/\t/g ;  1,2c SW score\t% div.\t% del.\t% ins.\tquery sequence\tpos in  query: begin\tend\t(left)\trepeat\tclass/family\tpos in repeat: begin\tend\t(left)\tID' rm_input.fasta.out >'${output_log}' &&
      mv rm_input.fasta.tbl '${output_table}' &&

        <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>

        <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>
    </conditional>
    <param type="boolean" argument="-gff" truevalue="-gff" falsevalue="" label="Output annotation of repeats in GFF format" checked="false" />
    <param argument="-excln" type="boolean" truevalue="-excln" falsevalue="" label="Ignore stretches of Ns when computing statistics" checked="true" help="Scaffolds are sometimes joined with stretches of 25 or more Ns. This option ignores them when calculating repeat statistics" />
    <section name="advanced" title="Advanced options" expanded="false">
      <param argument="-is_only" type="boolean" truevalue="-is_only" falsevalue="" checked="false" label="Only clip E coli insertion elements" />
      <param argument="-is_clip" type="boolean" truevalue="-is_clip" falsevalue="" checked="false" label="Clip IS elements before analysis" help="Normally RepeatMasker will report on IS element, with this option selected it will clip them before analysis" />
      <param argument="-no_is" type="boolean" truevalue="-no_is" falsevalue="" checked="false" label="Skip bacterial insertion element check" />
      <param argument="-rodspec" type="boolean" truevalue="-rodspec" falsevalue="" checked="false" label="Only check for rodent specific repeats" help="If this option is select a check for rodent specific repeats is done instead of a full RepeatMasker run" />

        <option value="-q">Quick (5-10% less sensitive, 3-4 times speedup)</option>
        <option value="-qq">Rush (10% less sensitive)</option>
        <option value="-s">Slow (0-5% more sensitive, 2.5 times slowdown)</option>
      </param>
      <param type="integer" argument="-frag" value="40000" label="Maximum contiguous sequence searched" help="Maximum length of sequencing that is search without fragmenting" />
      <!-- -maxsize option is in the help, but not in the code of repeatmasker-->
      <!--param type="integer" argument="-maxsize" value="4000000" label="Maximum length for IS or repeat clipped sequences" /-->
      <param type="integer" argument="-gc" optional="true" label="Select matrices for this GC%" help="Valid values are a percentage or -1 to choose the default" />
      <param type="boolean" argument="-gccalc" truevalue="-gcccalc" falsevalue="" checked="false" label="Calculate GC % for all sequences" help="By default RepeatMasker skips calculating GC % for small sequences" />
      <param type="boolean" argument="-nocut" truevalue="-nocut" falsevalue="" checked="false" label="Skips cutting of repeats" />
      <param name="xout" type="boolean" argument="-x" truevalue="-x" falsevalue="" checked="false" label="Mask with X instead of N characters" />
      <param name="keep_alignments" type="boolean" argument="-ali" truevalue="-ali" falsevalue="" checked="false" label="Output alignments file" />
      <param name="invert_alignments" type="boolean" argument="-inv" truevalue="-inv" falsevalue="" checked="false" label="Invert alignments in alignment file" help="Show alignments in the orientation of the repeat sequence, not the query sequence" />

      <output name="output_log" file="small.fasta.log" />
      <output name="output_alignment" file="small.fasta.align" />
      <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" />
      <output name="output_log" file="small_repbase.fasta.log" />
    </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
provided as a FASTA file or downloaded from RepBase_. If the RepBase option is