annotate variant_annotator.xml @ 4:f244b8209eb8 draft

bug fix release

<tool id="gatk2_variant_annotator" name="Variant Annotator" version="@VERSION@.0">
  <description></description>
  <expand macro="requirements" />
  <macros>
    <import>gatk2_macros.xml</import>
  </macros>
  <command interpreter="python">
    gatk2_wrapper.py
    --stdout "${output_log}"
    #if str( $reference_source.input_bam ) != "None":
        -d "-I" "${reference_source.input_bam}" "${reference_source.input_bam.ext}" "gatk_input"
        #if str( $reference_source.input_bam.metadata.bam_index ) != "None":
            -d "" "${reference_source.input_bam.metadata.bam_index}" "bam_index" "gatk_input" ##hardcode galaxy ext type as bam_index
        #end if
    #end if
    -d "--variant" "${reference_source.input_variant}" "${reference_source.input_variant.ext}" "input_variant"
    -p '
    @JAR_PATH@
    -T "VariantAnnotator"
    \$GATK2_SITE_OPTIONS

    @THREADS@

    #if $reference_source.reference_source_selector != "history":
        -R "${reference_source.ref_file.fields.path}"
    #end if
    -o "${output_vcf}"
    #if str( $annotations_type.annotations_type_selector ) == "use_all_annotations":
        --useAllAnnotations
    #else:
        #if $annotations_type.annotations:
            #for $annotation in str( $annotations_type.annotations.fields.gatk_value ).split( ',' ):
                --annotation "${annotation}"
            #end for
        #end if
    #end if
    #if $exclude_annotations:
        #for $annotation in str( $exclude_annotations.fields.gatk_value ).split( ',' ):
            --excludeAnnotation "${annotation}"
        #end for
    #end if
    #for $additional_annotation in $additional_annotations:
        --annotation "${additional_annotation.additional_annotation_name}"
    #end for
    '
    #if $reference_source.input_variant_bti:
        -d "--intervals" "${reference_source.input_variant}" "${reference_source.input_variant.ext}" "input_variant_bti"
    #end if
    
    #for $rod_binding in $comp_rod_bind:
        -d "--comp:${rod_binding.comp_rod_name},%(file_type)s" "${rod_binding.comp_input_rod}" "${rod_binding.comp_input_rod.ext}" "input_comp_${rod_binding.comp_rod_name}"
    #end for
    
    @DBSNP_OPTIONS@
    
    #for $rod_binding in $resource_rod_bind:
        -d "--resource:${rod_binding.resource_rod_name},%(file_type)s" "${rod_binding.resource_input_rod}" "${rod_binding.resource_input_rod.ext}" "input_resource_${rod_binding.resource_rod_name}"
    #end for
    
    #if str( $snpEff_rod_bind_type.snpEff_rod_bind_type_selector ) == 'set_snpEff':
        -p '--annotation "SnpEff"'
        -d "--snpEffFile:${snpEff_rod_bind_type.snpEff_rod_name},%(file_type)s" "${snpEff_rod_bind_type.snpEff_input_rod}" "${snpEff_rod_bind_type.snpEff_input_rod.ext}" "input_snpEff_${snpEff_rod_bind_type.snpEff_rod_name}"
    #else:
        -p '--excludeAnnotation "SnpEff"'
    #end if
    
    #for $expression in $expressions:
        -p '--expression "${expression.expression}"'
    #end for
    
    #include source=$standard_gatk_options#
    
    -p '
    #if str( $annotation_group ) != "None":
        #for $group in str( $annotation_group ).split( ',' ):
            --group "${group}"
        #end for
    #end if
    #if str( $family_string ) != "":
        --family_string "${family_string}"
    #end if
    --MendelViolationGenotypeQualityThreshold "${mendel_violation_genotype_quality_threshold}"
        '
  </command>
  <inputs>
    <conditional name="reference_source">
      <expand macro="reference_source_selector_param" />
      <when value="cached">
        <param name="input_variant" type="data" format="vcf" label="Variant file to annotate" help="-V,--variant &amp;lt;variant&amp;gt;"/>
        <param name="input_variant_bti" type="boolean" truevalue="-BTI variant" falsevalue="" label="Increase efficiency for small variant files." help="-BTI variant"/>
        <param name="input_bam" type="data" format="bam" label="BAM file" optional="True" help="Not needed for all annotations. (-I,--input_file &amp;lt;input_file&amp;gt;)" >
          <validator type="unspecified_build" />
          <validator type="dataset_metadata_in_data_table" table_name="gatk2_picard_indexes" metadata_name="dbkey" metadata_column="dbkey" message="Sequences are not currently available for the specified build." /> <!-- fixme!!! this needs to be a select -->
        </param>
        <param name="ref_file" type="select" label="Using reference genome" help="-R,--reference_sequence &amp;lt;reference_sequence&amp;gt;">
          <options from_data_table="gatk2_picard_indexes">
            <filter type="data_meta" key="dbkey" ref="input_variant" column="dbkey"/>
          </options>
          <validator type="no_options" message="A built-in reference genome is not available for the build associated with the selected input file"/>
        </param>
      </when>
      <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="input_variant_bti" type="boolean" truevalue="-BTI variant" falsevalue="" label="Increase efficiency for small variant files."  help="-BTI variant"/>
        <param name="input_bam" type="data" format="bam" label="BAM file" optional="True" help="Not needed for all annotations. (-I,--input_file &amp;lt;input_file&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>
    <conditional name="annotations_type">
      <param name="annotations_type_selector" type="select" label="Use all possible annotations">
        <option value="use_all_annotations">Use all</option>
        <option value="choose" selected="True">Use selected</option>
      </param>
      <when value="use_all_annotations">
          <!-- no extra options here -->
      </when>
      <when value="choose">
        <param name="annotations" type="select" multiple="True" display="checkboxes" label="Annotations to apply" help="-A,--annotation &amp;lt;annotation&amp;gt;" >
          <!-- load the available annotations from an external configuration file, since additional ones can be added to local installs -->
          <options from_data_table="gatk2_annotations">
            <filter type="multiple_splitter" column="tools_valid_for" separator=","/>
            <filter type="static_value" value="VariantAnnotator" column="tools_valid_for"/>
          </options>
        </param>
      </when>
    </conditional>
    
    <repeat name="additional_annotations" title="Additional annotation" help="-A,--annotation &amp;lt;annotation&amp;gt;">
      <param name="additional_annotation_name" type="text" value="" label="Annotation name" />
    </repeat>
    
    <repeat name="comp_rod_bind" title="Binding for reference-ordered comparison data" help="-comp,--comp &amp;lt;comp&amp;gt;">
      <param name="comp_input_rod" type="data" format="vcf" label="ROD file" />
      <param name="comp_rod_name" type="text" value="Unnamed" label="ROD Name"/>
    </repeat>
    <expand macro="dbsnp_param" />
    
    <repeat name="resource_rod_bind" title="Binding for reference-ordered resource data" help="-resource,--resource &amp;lt;resource&amp;gt;">
      <param name="resource_input_rod" type="data" format="vcf" label="ROD file" />
      <param name="resource_rod_name" type="text" value="Unnamed" label="ROD Name"/>
    </repeat>
    
    <conditional name="snpEff_rod_bind_type">
      <param name="snpEff_rod_bind_type_selector" type="select" label="Provide a snpEff reference-ordered data file (VCF)" help="-snpEffFile,--snpEffFile &amp;lt;snpEffFile&amp;gt;">
        <option value="set_snpEff">Set snpEff</option>
        <option value="exclude_snpEff" selected="True">Don't set snpEff</option>
      </param>
      <when value="exclude_snpEff">
        <!-- Do nothing here -->
      </when>
      <when value="set_snpEff">
        <param name="snpEff_input_rod" type="data" format="vcf" label="ROD file" />
        <param name="snpEff_rod_name" type="hidden" value="snpEff" label="ROD Name"/>
      </when>
    </conditional>
    
    <repeat name="expressions" title="Expression" help="-E,--expression &amp;lt;expression&amp;gt;">
      <param name="expression" type="text" value="" label="Expression"/>
    </repeat>
    
    <expand macro="gatk_param_type_conditional" />
    
    <param name="annotation_group" type="select" multiple="True" display="checkboxes" label="annotation interfaces/groups to apply to variant calls" help="-G,--group &amp;lt;group&amp;gt;">
      <option value="RodRequiringAnnotation">RodRequiringAnnotation</option>
      <option value="Standard">Standard</option>
      <option value="Experimental">Experimental</option>
      <option value="WorkInProgress">WorkInProgress</option>
      <option value="RankSumTest">RankSumTest</option>
    </param>
    <param name="family_string" type="text" value="" label="Family String" help="--family_string"/>
    <param name="mendel_violation_genotype_quality_threshold" type="float" value="0.0" label="genotype quality treshold in order to annotate mendelian violation ratio." help="-mvq,--MendelViolationGenotypeQualityThreshold &amp;lt;MendelViolationGenotypeQualityThreshold&amp;gt;"/>
    <param name="exclude_annotations" type="select" multiple="True" display="checkboxes" label="Annotations to exclude" help="-XA,--excludeAnnotation &amp;lt;excludeAnnotation&amp;gt;" >
      <!-- load the available annotations from an external configuration file, since additional ones can be added to local installs -->
      <options from_data_table="gatk2_annotations">
        <filter type="multiple_splitter" column="tools_valid_for" separator=","/>
        <filter type="static_value" value="VariantAnnotator" column="tools_valid_for"/>
      </options>
    </param>
    
  </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>
  <tests>
      <test>
          <param name="reference_source_selector" value="history" />
          <param name="ref_file" value="phiX.fasta" ftype="fasta" />
          <param name="input_bam" value="gatk/gatk_table_recalibration/gatk_table_recalibration_out_1.bam" ftype="bam" />
          <param name="input_variant" value="gatk/gatk_unified_genotyper/gatk_unified_genotyper_out_1.vcf" ftype="vcf" />
          <param name="input_variant_bti" />
          <param name="annotations_type_selector" value="choose" />
          <param name="annotations" value="AlleleBalance,BaseQualityRankSumTest,DepthOfCoverage,HomopolymerRun,MappingQualityRankSumTest,MappingQualityZero,QualByDepth,RMSMappingQuality,SpanningDeletions,HaplotypeScore" />
          <param name="additional_annotations" value="0" />
          <param name="dbsnp_rod_bind_type_selector" value="set_dbsnp" />
          <param name="dbsnp_input_rod" value="gatk/fake_phiX_variant_locations.vcf" ftype="vcf" />
          <param name="dbsnp_rod_name" value="dbsnp" />
          <param name="snpEff_rod_bind_type_selector" value="exclude_snpEff" />
          <param name="gatk_param_type_selector" value="basic" />
          <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_annotator/gatk_variant_annotator_out_1.log.contains" compare="contains" />
          <param name="comp_rod_bind" value="0" />
          <param name="resource_rod_bind" value="0" />
          <param name="expressions" value="0" />
          <!-- <param name="annotation_group" /> -->
      </test>
  </tests>
  <help>
**What it does**

Annotates variant calls with context information.  Users can specify which of the available annotations to use.

For more information on using the VariantAnnotator, see this `tool specific page &lt;http://www.broadinstitute.org/gatk/gatkdocs/org_broadinstitute_sting_gatk_walkers_annotator_VariantAnnotator.html&gt;`_.

To learn about best practices for variant detection using GATK, see this `overview &lt;http://www.broadinstitute.org/gatk/guide/topic?name=best-practices&gt;`_.

If you encounter errors, please view the `GATK FAQ &lt;http://www.broadinstitute.org/gatk/guide/topic?name=faqs&gt;`_.

------


**Inputs**

GenomeAnalysisTK: VariantAnnotator accepts a variant input file.


**Outputs**

The output is in VCF format.


Go `here &lt;http://www.broadinstitute.org/gatk/guide/topic?name=intro&gt;`_ for details on GATK file formats.

-------

**Settings**::


 sampleName           The sample (NA-ID) corresponding to the variant input (for non-VCF input only)
 annotation           One or more specific annotations to apply to variant calls
 group                One or more classes/groups of annotations to apply to variant calls
 expression           One or more specific expressions to apply to variant calls; see documentation for more details
 useAllAnnotations    Use all possible annotations (not for the faint of heart)

@CITATION_SECTION@
  </help>
  <expand macro="citations" />
</tool>