comparison microsatbed.xml @ 20:410144c7b2d6 draft

planemo upload for repository https://github.com/fubar2/microsatbed commit d952bc313f408735456747c3d33e09a3170c8f59-dirty

19:db5523378e5c

  <description>Short Tandem Repeats to bed features from fasta</description>
  <requirements>
    <requirement version="3.12.3" type="package">python</requirement>
    <requirement version="2.1.0" type="package">pyfastx</requirement>
    <requirement version="1.3.0" type="package">pytrf</requirement>
  </requirements>
   <required_files>
      <include path="find_str.py"/>
  </required_files>
   <version_command><![CDATA[python -c "import pytrf; from importlib.metadata import version; print(version('pytrf'))"]]></version_command>

      --fasta '${reference_genome.fasta.fields.path}'
    #end if
    --bed '$bed'
    #if $mode_cond.mode == "SPECIFIC":
    --specific '$mode_cond.specific'
    #else:
      #if "MONO" in $mode_cond.subset:
      --mono
      #end if
      #if "DI" in $mode_cond.subset:

    --dimin '$dimin'
    --trimin '$trimin'
    --tetramin '$tetramin'
    --pentamin '$pentamin'
    --hexamin '$hexamin'
  #end if
]]></command>
  <inputs>
    <conditional name="reference_genome">
        <param name="genome_type_select" type="select" label="Select a source for fasta sequences to be searched for STRs" help="Options are to choose a built-in genome, or choose any history fasta file">

        </when>
    </conditional>
    <conditional name="mode_cond">
    <param name="mode" type="select" label="Select patterns by motif length; or provide a specific motif pattern to report?" help="Choose *By length:* or *By pattern:* to configure STR selection mode">
      <option selected="True" value="ALL">By length: Report all motifs of one or more specified lengths (1-6nt) as bed features</option>
      <option value="SPECIFIC">By motif: Report one or more specific motifs (such as TCA,GC) as bed features</option>
      <option value="NATIVE">All exact STR: use the pytrf findstr native command to a create csv, tsv or gtf output</option>
    </param>
    <when value="ALL">
      <param name="subset" type="select" multiple="true" optional="false" label="Select at least 1 specific motif length to report" 
        help="Bed features will be output for every motif of the selected length(s) with the minimum required repeats or more">
      <option value="DI" selected="true">All dimers (AC,AG,AT,...)</option>
      <option value="TRI">All trimers (ACG,..)</option>                                                                                                                          
      <option value="TETRA">All tetramers (ACGT,..)</option>
      <option value="PENTA">All pentamers (ACGTC,..)</option>
      <option value="HEXA">All hexamers (ACGTCG,..)</option>
      <option value="MONO">All monomers (A,C...). Warning! Can produce overwhelming numbers of bed features</option>
      </param>
    </when>
    <when value="SPECIFIC">
      <param name="specific" type="text" label="Supply a specific motif pattern. Separate multiple patterns with commas such as GA,GC" 
         help="Make bed features only for the nominated specific motifs." optional="false"/>
    </when>
    <when value="NATIVE">
      <param name="outformat" type="select" optional="false" label="Select the required output format" 
        help="Pytrf can create GFF, CSV or TSV output files. Documentation is linked in the help section below">
        <option value="gff" >GFF</option>

    <data name="bed" format="bed" label="STR from $fasta.element_identifier" hidden="false">
      <change_format>
            <when input="mode_cond.outformat" value="gff" format="gff" />
            <when input="mode_cond.outformat" value="csv" format="csv" />
            <when input="mode_cond.outformat" value="tsv" format="tabular" />
      </change_format>
    </data>
  </outputs>
  <tests>
    <test expect_num_outputs="1">

      <param name="trimin" value="5"/>
      <param name="tetramin" value="4"/>
      <param name="pentamin" value="4"/>
      <param name="hexamin" value="2"/>
      <output name="bed" value="nativegff_sample" compare="diff" lines_diff="0"/>
    </test>
  </tests>
  <help><![CDATA[

 **Convert short repetitive sequences to bed features**

20:410144c7b2d6

  <description>Short Tandem Repeats to bed features from fasta</description>
  <requirements>
    <requirement version="3.12.3" type="package">python</requirement>
    <requirement version="2.1.0" type="package">pyfastx</requirement>
    <requirement version="1.3.0" type="package">pytrf</requirement>
    <requirement type="package" version="0.1.4">pybigtools</requirement>
  </requirements>
   <required_files>
      <include path="find_str.py"/>
  </required_files>
   <version_command><![CDATA[python -c "import pytrf; from importlib.metadata import version; print(version('pytrf'))"]]></version_command>

      --fasta '${reference_genome.fasta.fields.path}'
    #end if
    --bed '$bed'
    #if $mode_cond.mode == "SPECIFIC":
    --specific '$mode_cond.specific'
    #elif $mode_cond.mode == "SPECIFICBW":
    --bigwig
    --winwidth '$mode_cond.winwidth'
    --specific '$mode_cond.specific'
    #else:
      #if "MONO" in $mode_cond.subset:
      --mono
      #end if
      #if "DI" in $mode_cond.subset:

    --dimin '$dimin'
    --trimin '$trimin'
    --tetramin '$tetramin'
    --pentamin '$pentamin'
    --hexamin '$hexamin'
    #if $mode_cond.mode == "SPECIFICBW":
    --bigwig
    --winwidth '$mode_cond.winwidth'
    #end if
  #end if
]]></command>
  <inputs>
    <conditional name="reference_genome">
        <param name="genome_type_select" type="select" label="Select a source for fasta sequences to be searched for STRs" help="Options are to choose a built-in genome, or choose any history fasta file">

        </when>
    </conditional>
    <conditional name="mode_cond">
    <param name="mode" type="select" label="Select patterns by motif length; or provide a specific motif pattern to report?" help="Choose *By length:* or *By pattern:* to configure STR selection mode">
      <option selected="True" value="ALL">By length: Report all motifs of one or more specified lengths (1-6nt) as bed features</option>
      <option value="ALLBW">By length as windowed bigwig: Report all motifs of one or more specified lengths (1-6nt) as windowed density</option>
      <option value="SPECIFIC">By motif: Report one or more specific motifs (such as TCA,GC) as bed features</option>
      <option value="SPECIFICBW">By motif as windowed bigwig: Report one or more specific motifs (such as TCA,GC) as windowed density</option>
      <option value="NATIVE">All exact STR: use the pytrf findstr native command to a create csv, tsv or gtf output</option>
    </param>
    <when value="ALL">
      <param name="subset" type="select" multiple="true" optional="false" label="Select at least 1 specific motif length to report" 
        help="Bed features will be output for every motif of the selected length(s) with the minimum required repeats or more">
      <option value="DI" selected="true">All dimers (AC,AG,AT,...)</option>
      <option value="TRI">All trimers (ACG,..)</option>
      <option value="PENTA">All pentamers (ACGTC,..)</option>
      <option value="HEXA">All hexamers (ACGTCG,..)</option>
      <option value="MONO">All monomers (A,C...). Warning! Can produce overwhelming numbers of bed features</option>
      </param>
    </when>
    <when value="ALLBW">
      <param name="subset" type="select" multiple="true" optional="false" label="Select at least 1 specific motif length to report" 
        help="Bed features will be output for every motif of the selected length(s) with the minimum required repeats or more">
      <option value="DI" selected="true">All dimers (AC,AG,AT,...)</option>
      <option value="TRI">All trimers (ACG,..)</option>
      <option value="PENTA">All pentamers (ACGTC,..)</option>
      <option value="HEXA">All hexamers (ACGTCG,..)</option>
      <option value="MONO">All monomers (A,C...). Warning! Can produce overwhelming numbers of bed features</option>
      </param>
      <param type="integer" min="5" name="winwidth" label="Window with for estimating STR bigwig density" value="128"/>
    </when>
    <when value="SPECIFIC">
      <param name="specific" type="text" label="Supply a specific motif pattern. Separate multiple patterns with commas such as GA,GC" 
         help="Make bed features only for the nominated specific motifs." optional="false"/>
    </when>
    <when value="SPECIFICBW">
      <param name="specific" type="text" label="Supply a specific motif pattern. Separate multiple patterns with commas such as GA,GC" 
         help="Make bed features only for the nominated specific motifs." optional="false"/>
      <param type="integer" min="5" name="winwidth" label="Window with for estimating STR bigwig density" value="128"/>
    </when>
    <when value="NATIVE">
      <param name="outformat" type="select" optional="false" label="Select the required output format" 
        help="Pytrf can create GFF, CSV or TSV output files. Documentation is linked in the help section below">
        <option value="gff" >GFF</option>

    <data name="bed" format="bed" label="STR from $fasta.element_identifier" hidden="false">
      <change_format>
            <when input="mode_cond.outformat" value="gff" format="gff" />
            <when input="mode_cond.outformat" value="csv" format="csv" />
            <when input="mode_cond.outformat" value="tsv" format="tabular" />
            <when input="mode_cond.mode" value="ALLBW" format="bigwig" />
            <when input="mode_cond.mode" value="SPECIFICBW" format="bigwig" />
      </change_format>
    </data>
  </outputs>
  <tests>
    <test expect_num_outputs="1">

      <param name="trimin" value="5"/>
      <param name="tetramin" value="4"/>
      <param name="pentamin" value="4"/>
      <param name="hexamin" value="2"/>
      <output name="bed" value="nativegff_sample" compare="diff" lines_diff="0"/>
    </test>
    <test expect_num_outputs="1">
      <conditional name="reference_genome">
          <param name="genome_type_select" value="history"/>
          <param name="fasta" value="humsamp.fa"/>
      </conditional>
      <conditional name="mode_cond">
          <param name="mode" value="SPECIFICBW"/>
          <param name="specific" value="GC"/>
      </conditional>
      <param name="monomin" value="20"/>
      <param name="dimin" value="1"/>
      <param name="trimin" value="20"/>
      <param name="tetramin" value="20"/>
      <param name="pentamin" value="20"/>
      <param name="hexamin" value="20"/>
      <output name="bed" value="dibed_wig_sample" compare="sim_size" delta="10"/>
    </test>
  </tests>
  <help><![CDATA[

 **Convert short repetitive sequences to bed features**