diff unified_genotyper.xml @ 2:8bcc13094767 draft

Uploaded
author iuc
date Sat, 18 Jan 2014 07:21:33 -0500
parents 340633249b3d
children 2553f84b8174
line wrap: on
line diff
--- a/unified_genotyper.xml	Mon Dec 02 10:36:02 2013 -0500
+++ b/unified_genotyper.xml	Sat Jan 18 07:21:33 2014 -0500
@@ -31,22 +31,12 @@
     --standard_min_confidence_threshold_for_calling "${standard_min_confidence_threshold_for_calling}"
     --standard_min_confidence_threshold_for_emitting "${standard_min_confidence_threshold_for_emitting}"
    '
-    #set $rod_binding_names = dict()
-    #for $rod_binding in $rod_bind:
-        #if str( $rod_binding.rod_bind_type.rod_bind_type_selector ) == 'custom':
-            #set $rod_bind_name = $rod_binding.rod_bind_type.custom_rod_name
-        #else
-            #set $rod_bind_name = $rod_binding.rod_bind_type.rod_bind_type_selector
-        #end if
-        #set $rod_binding_names[$rod_bind_name] = $rod_binding_names.get( $rod_bind_name, -1 ) + 1
-        -d "--dbsnp:${rod_bind_name},%(file_type)s" "${rod_binding.rod_bind_type.input_rod}" "${rod_binding.rod_bind_type.input_rod.ext}" "input_${rod_bind_name}_${rod_binding_names[$rod_bind_name]}"
-    #end for
-   
+    @DBSNP_OPTIONS@
+
     #include source=$standard_gatk_options#
     ##start analysis specific options
     #if $analysis_param_type.analysis_param_type_selector == "advanced":
         -p '
-        --p_nonref_model "${analysis_param_type.p_nonref_model}"
         --heterozygosity "${analysis_param_type.heterozygosity}"
         --pcr_error_rate "${analysis_param_type.pcr_error_rate}"
         --genotyping_mode "${analysis_param_type.genotyping_mode_type.genotyping_mode}"
@@ -120,30 +110,7 @@
         <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="rod_bind" title="Binding for reference-ordered data" help="-D,--dbsnp &amp;lt;dbsnp&amp;gt;">
-        <conditional name="rod_bind_type">
-          <param name="rod_bind_type_selector" type="select" label="Binding Type">
-            <option value="dbsnp" selected="True">dbSNP</option>
-            <option value="snps">SNPs</option>
-            <option value="indels">INDELs</option>
-            <option value="custom">Custom</option>
-          </param>
-          <when value="dbsnp">
-              <param name="input_rod" type="data" format="vcf" label="ROD file" />
-          </when>
-          <when value="snps">
-              <param name="input_rod" type="data" format="vcf" label="ROD file" />
-          </when>
-          <when value="indels">
-              <param name="input_rod" type="data" format="vcf" label="ROD file" />
-          </when>
-          <when value="custom">
-              <param name="custom_rod_name" type="text" value="Unknown" label="ROD Name"/>
-              <param name="input_rod" type="data" format="vcf" label="ROD file" />
-          </when>
-        </conditional>
-    </repeat>
+    <expand macro="dbsnp_param" />
     
     <param name="genotype_likelihoods_model" type="select" label="Genotype likelihoods calculation model to employ" help="-glm,--genotype_likelihoods_model &amp;lt;genotype_likelihoods_model&amp;gt;">
       <option value="BOTH" selected="True">BOTH</option>
@@ -158,10 +125,6 @@
     <expand macro="gatk_param_type_conditional" />
     
     <expand macro="analysis_type_conditional">
-        <param name="p_nonref_model" type="select" label="Non-reference probability calculation model to employ" help="-pnrm,--p_nonref_model &amp;lt;p_nonref_model&amp;gt;">
-          <option value="EXACT_GENERAL_PLOIDY" selected="True">EXACT_GENERAL_PLOIDY (supports any sample ploidy)</option>
-          <option value="EXACT_REFERENCE">EXACT_REFERENCE (multi-allelic EXACT model.  Extremely slow for many alternate alleles)</option>
-        </param>
         <param name="heterozygosity" type="float" value="1e-3" label="Heterozygosity value used to compute prior likelihoods for any locus" help="-hets,--heterozygosity &amp;lt;heterozygosity&amp;gt;" />
         <param name="pcr_error_rate" type="float" value="1e-4" label="The PCR error rate to be used for computing fragment-based likelihoods" help="-pcr_error,--pcr_error_rate &amp;lt;pcr_error_rate&amp;gt;" />
         <conditional name="genotyping_mode_type">
@@ -187,13 +150,11 @@
         <param name="max_alternate_alleles" type="integer" value="6" label="Maximum number of alternate alleles to genotype" help="-maxAlleles,--max_alternate_alleles &amp;lt;max_alternate_alleles&amp;gt;" />
         <param name="min_indel_count_for_genotyping" type="integer" value="5" label="Minimum number of consensus indels required to trigger genotyping run" help="-minIndelCnt,--min_indel_count_for_genotyping &amp;lt;min_indel_count_for_genotyping&amp;gt;" />
         <param name="indel_heterozygosity" type="float" value="0.000125" label="Heterozygosity for indel calling" help="1.0/8000==0.000125 (-indelHeterozygosity,--indel_heterozygosity &amp;lt;indel_heterozygosity&amp;gt;)"/>
-        <param name="indelGapContinuationPenalty" type="integer" value="10" label="Indel gap continuation penalty" help="--indelGapContinuationPenalty">
-          <help>Indel gap continuation penalty, as Phred-scaled probability. I.e., 30 => 10^-30/10.</help>
-          <validator type="in_range" message="value between 0 and 255" min="0" max="255"/> 
+        <param name="indelGapContinuationPenalty" type="integer" value="10" label="Indel gap continuation penalty" help="As Phred-scaled probability, i.e. 30 => 10^-30/10 (--indelGapContinuationPenalty)">
+          <validator type="in_range" message="value between 0 and 255" min="0" max="255" />
         </param>
-        <param name="indelGapOpenPenalty" type="integer" value="45" label="Indel gap open penalty" help="--indelGapOpenPenalty" >
-          <help>Indel gap open penalty, as Phred-scaled probability. I.e., 30 => 10^-30/10.</help>
-          <validator type="in_range" message="value between 0 and 255" min="0" max="255"/> 
+        <param name="indelGapOpenPenalty" type="integer" value="45" label="Indel gap open penalty" help="As Phred-scaled probability, i.e. 30 => 10^-30/10 (--indelGapOpenPenalty)">
+          <validator type="in_range" message="value between 0 and 255" min="0" max="255" />
         </param>
         <!-- indelHaplotypeSize - Gone in GATK 2.4? -->
         <param name="indelHaplotypeSize" type="integer" value="80" label="Indel haplotype size" help="--indelHaplotypeSize" />
@@ -252,14 +213,14 @@
           <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="rod_bind_type_selector" value="dbsnp" />
-          <param name="input_rod" value="gatk/fake_phiX_variant_locations.vcf" ftype="vcf" />
+          <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="standard_min_confidence_threshold_for_calling" value="0" />
           <param name="standard_min_confidence_threshold_for_emitting" value="4" />
           <param name="gatk_param_type_selector" value="basic" />
           <param name="analysis_param_type_selector" value="advanced" />
           <param name="genotype_likelihoods_model" value="BOTH" />
-          <param name="p_nonref_model" value="EXACT" />
           <param name="heterozygosity" value="0.001" />
           <param name="pcr_error_rate" value="0.0001" />
           <param name="genotyping_mode" value="DISCOVERY" />
@@ -310,7 +271,6 @@
 **Settings**::
 
  genotype_likelihoods_model                        Genotype likelihoods calculation model to employ -- BOTH is the default option, while INDEL is also available for calling indels and SNP is available for calling SNPs only (SNP|INDEL|BOTH)
- p_nonref_model                                    Non-reference probability calculation model to employ -- EXACT_GENERAL_PLOIDY is the default option, while EXACT_REFERENCE is also available. (EXACT_INDEPENDENT,EXACT_REFERENCE,EXACT_ORIGINAL,EXACT_GENERAL_PLOIDY)
  heterozygosity                                    Heterozygosity value used to compute prior likelihoods for any locus
  pcr_error_rate                                    The PCR error rate to be used for computing fragment-based likelihoods
  genotyping_mode                                   Should we output confident genotypes (i.e. including ref calls) or just the variants? (DISCOVERY|GENOTYPE_GIVEN_ALLELES)