comparison variant_filtration.xml @ 6:35c00763cb5c draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/gatk2 commit cf399638ebca4250bcc15f468238a9964de97b33

5:84584664264c

<tool id="gatk2_variant_filtration" name="Variant Filtration" version="@VERSION@.0">
  <description>on VCF files</description>
  <expand macro="requirements" />
  <macros>
    <import>gatk2_macros.xml</import>
  </macros>
  <command interpreter="python">
    #from binascii import hexlify

    gatk2_wrapper.py
    --stdout "${output_log}"

    '
    #for $variant_filter in $variant_filters:
        #set $variant_filter = "--%sExpression '%s' --%sName '%s'" % ( str( $variant_filter.is_genotype_filter ), str( $variant_filter.filter_expression ), str( $variant_filter.is_genotype_filter ), str( $variant_filter.filter_name )  )
        -o '${ hexlify( $variant_filter ) }'
    #end for
    
    #if str( $mask_rod_bind_type.mask_rod_bind_type_selector ) == 'set_mask':
        -d "--mask:${mask_rod_bind_type.mask_rod_name},%(file_type)s" "${mask_rod_bind_type.input_mask_rod}" "${mask_rod_bind_type.input_mask_rod.ext}" "input_mask_${mask_rod_bind_type.mask_rod_name}"
        -p '
        --maskExtension "${mask_rod_bind_type.mask_extension}"
        --maskName "${mask_rod_bind_type.mask_rod_name}"
        '
    #end if
    
    #include source=$standard_gatk_options#
    
    ##start analysis specific options
    #if $cluster_snp_type.cluster_snp_type_selector == "cluster_snp":
        -p '
        --clusterSize "${cluster_snp_type.cluster_size}"
        --clusterWindowSize "${cluster_snp_type.cluster_window_size}"

      <when value="history"> <!-- FIX ME!!!! -->
        <param name="input_variant" type="data" format="vcf" label="Variant file to annotate" help="-V,--variant &amp;lt;variant&amp;gt;" />
        <param name="ref_file" type="data" format="fasta" label="Using reference file" help="-R,--reference_sequence &amp;lt;reference_sequence&amp;gt;" />
      </when>
    </conditional>
    
    
    <repeat name="variant_filters" title="Variant Filters">
        <param name="filter_expression" value="AB &lt; 0.2 || MQ0 &gt; 50" type="text" label="Filter expression" help="JEXL formatted expressions (-filter,--filterExpression &amp;lt;filterExpression&amp;gt;)">
            <sanitizer>
              <valid initial="string.printable">
               <remove value="&apos;"/>

            </sanitizer>
        </param>
        <param name="filter_name" value="custom_filter" type="text" label="Filter name" help="-filterName,--filterName &amp;lt;filterName&amp;gt;"/>
        <param name="is_genotype_filter" type="boolean" truevalue="genotypeFilter" falsevalue="filter" label="Use filter at the individual sample level" help="Use -G_filter,--genotypeFilterExpression &amp;lt;genotypeFilterExpression&amp;gt; and -G_filterName,--genotypeFilterName &amp;lt;genotypeFilterName&amp;gt; for filter type" />
    </repeat>
    

    
    <conditional name="mask_rod_bind_type">
      <param name="mask_rod_bind_type_selector" type="select" label="Provide a Mask reference-ordered data file">
        <option value="set_mask" selected="True">Set mask</option>
        <option value="exclude_mask">Don't set mask</option>
      </param>

      <when value="set_mask">
        <param name="input_mask_rod" type="data" format="bed,gatk_dbsnp,vcf" label="Mask ROD file" help="--mask &amp;lt;mask&amp;gt;" />
        <param name="mask_rod_name" type="text" value="Mask" label="Mask Name" help="-maskName,--maskName &amp;lt;maskName&amp;gt;"/>
        <param name="mask_extension" type="integer" value="0" label="Mask Extension" help="-maskExtend,--maskExtension &amp;lt;maskExtension&amp;gt;"/>
      </when>
    </conditional>    
    
    
    <expand macro="gatk_param_type_conditional" />
    
    <conditional name="cluster_snp_type">
      <param name="cluster_snp_type_selector" type="select" label="Cluster SNPs">
        <option value="cluster_snp">Cluster SNPs</option>
        <option value="do_not_cluster_snp" selected="True">Do not cluster SNPs</option>
      </param>

      <when value="cluster_snp">
        <param name="cluster_size" type="integer" value="3" label="The number of SNPs which make up a cluster" help="-cluster,--clusterSize &amp;lt;clusterSize&amp;gt;"/>
        <param name="cluster_window_size" type="integer" value="0" label="The window size (in bases) in which to evaluate clustered SNPs" help="-window,--clusterWindowSize &amp;lt;clusterWindowSize&amp;gt;"/>
      </when>
    </conditional>
    
    <param name="missing_values_in_expressions_should_evaluate_as_failing" type="boolean" truevalue="--missingValuesInExpressionsShouldEvaluateAsFailing" falsevalue="" label="Should missing values be considered failing the expression" help="--missingValuesInExpressionsShouldEvaluateAsFailing" />
    
  </inputs>
  <outputs>
    <data format="vcf" name="output_vcf" label="${tool.name} on ${on_string} (Variant File)" />
    <data format="txt" name="output_log" label="${tool.name} on ${on_string} (log)" />
  </outputs>

          <param name="mask_rod_name" value="." />
          <param name="mask_extension" value="0" />
          <param name="gatk_param_type_selector" value="basic" />
          <param name="cluster_snp_type_selector" value="do_not_cluster_snp" />
          <param name="missing_values_in_expressions_should_evaluate_as_failing" />
          <output name="output_vcf" file="gatk/gatk_variant_annotator/gatk_variant_annotator_out_1.vcf" lines_diff="4" /> 
          <output name="output_log" file="gatk/gatk_variant_filtration/gatk_variant_filtration_out_1.log.contains" compare="contains" />
      </test>
  </tests>
  <help>
**What it does**

6:35c00763cb5c

<tool id="gatk2_variant_filtration" name="Variant Filtration" version="@VERSION@.0">
  <description>on VCF files</description>
  <macros>
    <import>gatk2_macros.xml</import>
  </macros>
  <expand macro="requirements" />
  <expand macro="version_command" />
  <command interpreter="python">
    #from binascii import hexlify

    gatk2_wrapper.py
    --stdout "${output_log}"

    '
    #for $variant_filter in $variant_filters:
        #set $variant_filter = "--%sExpression '%s' --%sName '%s'" % ( str( $variant_filter.is_genotype_filter ), str( $variant_filter.filter_expression ), str( $variant_filter.is_genotype_filter ), str( $variant_filter.filter_name )  )
        -o '${ hexlify( $variant_filter ) }'
    #end for

    #if str( $mask_rod_bind_type.mask_rod_bind_type_selector ) == 'set_mask':
        -d "--mask:${mask_rod_bind_type.mask_rod_name},%(file_type)s" "${mask_rod_bind_type.input_mask_rod}" "${mask_rod_bind_type.input_mask_rod.ext}" "input_mask_${mask_rod_bind_type.mask_rod_name}"
        -p '
        --maskExtension "${mask_rod_bind_type.mask_extension}"
        --maskName "${mask_rod_bind_type.mask_rod_name}"
        '
    #end if

    #include source=$standard_gatk_options#

    ##start analysis specific options
    #if $cluster_snp_type.cluster_snp_type_selector == "cluster_snp":
        -p '
        --clusterSize "${cluster_snp_type.cluster_size}"
        --clusterWindowSize "${cluster_snp_type.cluster_window_size}"

      <when value="history"> <!-- FIX ME!!!! -->
        <param name="input_variant" type="data" format="vcf" label="Variant file to annotate" help="-V,--variant &amp;lt;variant&amp;gt;" />
        <param name="ref_file" type="data" format="fasta" label="Using reference file" help="-R,--reference_sequence &amp;lt;reference_sequence&amp;gt;" />
      </when>
    </conditional>

    <repeat name="variant_filters" title="Variant Filters">
        <param name="filter_expression" value="AB &lt; 0.2 || MQ0 &gt; 50" type="text" label="Filter expression" help="JEXL formatted expressions (-filter,--filterExpression &amp;lt;filterExpression&amp;gt;)">
            <sanitizer>
              <valid initial="string.printable">
               <remove value="&apos;"/>

            </sanitizer>
        </param>
        <param name="filter_name" value="custom_filter" type="text" label="Filter name" help="-filterName,--filterName &amp;lt;filterName&amp;gt;"/>
        <param name="is_genotype_filter" type="boolean" truevalue="genotypeFilter" falsevalue="filter" label="Use filter at the individual sample level" help="Use -G_filter,--genotypeFilterExpression &amp;lt;genotypeFilterExpression&amp;gt; and -G_filterName,--genotypeFilterName &amp;lt;genotypeFilterName&amp;gt; for filter type" />
    </repeat>

    <conditional name="mask_rod_bind_type">
      <param name="mask_rod_bind_type_selector" type="select" label="Provide a Mask reference-ordered data file">
        <option value="set_mask" selected="True">Set mask</option>
        <option value="exclude_mask">Don't set mask</option>
      </param>

      <when value="set_mask">
        <param name="input_mask_rod" type="data" format="bed,gatk_dbsnp,vcf" label="Mask ROD file" help="--mask &amp;lt;mask&amp;gt;" />
        <param name="mask_rod_name" type="text" value="Mask" label="Mask Name" help="-maskName,--maskName &amp;lt;maskName&amp;gt;"/>
        <param name="mask_extension" type="integer" value="0" label="Mask Extension" help="-maskExtend,--maskExtension &amp;lt;maskExtension&amp;gt;"/>
      </when>
    </conditional>

    <expand macro="gatk_param_type_conditional" />

    <conditional name="cluster_snp_type">
      <param name="cluster_snp_type_selector" type="select" label="Cluster SNPs">
        <option value="cluster_snp">Cluster SNPs</option>
        <option value="do_not_cluster_snp" selected="True">Do not cluster SNPs</option>
      </param>

      <when value="cluster_snp">
        <param name="cluster_size" type="integer" value="3" label="The number of SNPs which make up a cluster" help="-cluster,--clusterSize &amp;lt;clusterSize&amp;gt;"/>
        <param name="cluster_window_size" type="integer" value="0" label="The window size (in bases) in which to evaluate clustered SNPs" help="-window,--clusterWindowSize &amp;lt;clusterWindowSize&amp;gt;"/>
      </when>
    </conditional>

    <param name="missing_values_in_expressions_should_evaluate_as_failing" type="boolean" truevalue="--missingValuesInExpressionsShouldEvaluateAsFailing" falsevalue="" label="Should missing values be considered failing the expression" help="--missingValuesInExpressionsShouldEvaluateAsFailing" />

  </inputs>
  <outputs>
    <data format="vcf" name="output_vcf" label="${tool.name} on ${on_string} (Variant File)" />
    <data format="txt" name="output_log" label="${tool.name} on ${on_string} (log)" />
  </outputs>

          <param name="mask_rod_name" value="." />
          <param name="mask_extension" value="0" />
          <param name="gatk_param_type_selector" value="basic" />
          <param name="cluster_snp_type_selector" value="do_not_cluster_snp" />
          <param name="missing_values_in_expressions_should_evaluate_as_failing" />
          <output name="output_vcf" file="gatk/gatk_variant_annotator/gatk_variant_annotator_out_1.vcf" lines_diff="4" />
          <output name="output_log" file="gatk/gatk_variant_filtration/gatk_variant_filtration_out_1.log.contains" compare="contains" />
      </test>
  </tests>
  <help>
**What it does**