Mercurial > repos > iuc > gatk2

comparison haplotype_caller.xml @ 0:340633249b3d draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author bgruening
date Mon, 02 Dec 2013 06:18:36 -0500
parents
children 8bcc13094767
comparison
equal deleted inserted replaced
-1:000000000000 0:340633249b3d
1 <tool id="gatk2_haplotype_caller" name="Haplotype Caller" version="0.0.7">
2 <description>Call SNPs and indels simultaneously via local de-novo assembly of haplotypes in an active region</description>
3 <expand macro="requirements" />
4 <macros>
5 <import>gatk2_macros.xml</import>
6 </macros>
7 <command interpreter="python">
8 gatk2_wrapper.py
9 --stdout "${output_log}"
10 -d "-I" "${reference_source.input_bam}" "${reference_source.input_bam.ext}" "gatk_input"
11 #if str( $reference_source.input_bam.metadata.bam_index ) != "None":
12 -d "" "${reference_source.input_bam.metadata.bam_index}" "bam_index" "gatk_input" ##hardcode galaxy ext type as bam_index
13 #end if
14 -p '
15 @JAR_PATH@
16 -T "HaplotypeCaller"
17 -o "${output_vcf}"
18
19 \$GATK2_SITE_OPTIONS
20
21 @THREADS@
22
23 #if $reference_source.reference_source_selector != "history":
24 -R "${reference_source.ref_file.fields.path}"
25 #end if
26 #if str($input_recal) != 'None':
27 --BQSR "${input_recal}"
28 #end if
29 '
30 #include source=$standard_gatk_options#
31
32 ##start analysis specific options
33 #if $analysis_param_type.analysis_param_type_selector == "advanced":
34 -p '
35 #if $analysis_param_type.p_nonref_model.__str__ != "None" and len($analysis_param_type.p_nonref_model.__str__) > 0:
36 --p_nonref_model $analysis_param_type.p_nonref_model
37 #end if
38 #if $analysis_param_type.heterozygosity.__str__.strip() != '':
39 --heterozygosity $analysis_param_type.heterozygosity
40 #end if
41 --genotyping_mode "${analysis_param_type.genotyping_mode_type.genotyping_mode}"
42 #if str( $analysis_param_type.genotyping_mode_type.genotyping_mode ) == 'GENOTYPE_GIVEN_ALLELES':
43 --alleles "${analysis_param_type.genotyping_mode_type.input_alleles_rod}"
44 #end if
45 #if $analysis_param_type.output_mode.__str__ != "None" and len($analysis_param_type.output_mode.__str__) > 0:
46 --output_mode $analysis_param_type.output_mode
47 #end if
48
49 ## files
50 #if str($analysis_param_type.activeRegionIn) != 'None':
51 --activeRegionIn "$analysis_param_type.activeRegionIn"
52 #end if
53 #if str($analysis_param_type.comp) != 'None':
54 --comp "$analysis_param_type.comp"
55 #end if
56 #if str($analysis_param_type.dbsnp) != 'None':
57 --dbsnp "$analysis_param_type.dbsnp"
58 #end if
59 ##
60 #if str( $analysis_param_type.annotation ) != "None":
61 #for $annotation in str( $analysis_param_type.annotation.fields.gatk_value ).split( ','):
62 --annotation "${annotation}"
63 #end for
64 #end if
65 #for $additional_annotation in $analysis_param_type.additional_annotations:
66 --annotation "${additional_annotation.additional_annotation_name}"
67 #end for
68 #if str( $analysis_param_type.group ) != "None":
69 #for $group in str( $analysis_param_type.group ).split( ','):
70 --group "${group}"
71 #end for
72 #end if
73 #if str( $analysis_param_type.exclude_annotations ) != "None":
74 #for $annotation in str( $analysis_param_type.exclude_annotations.fields.gatk_value ).split( ','):
75 --excludeAnnotation "${annotation}"
76 #end for
77 #end if
78
79 ## value setings
80 #if $analysis_param_type.contamination_fraction_to_filter.__str__.strip() != '':
81 --contamination_fraction_to_filter $analysis_param_type.contamination_fraction_to_filter
82 #end if
83 #if $analysis_param_type.downsampleRegion.__str__.strip() != '':
84 --downsampleRegion $analysis_param_type.downsampleRegion
85 #end if
86 #if $analysis_param_type.minPruning.__str__.strip() != '':
87 --minPruning $analysis_param_type.minPruning
88 #end if
89 #if $analysis_param_type.standard_min_confidence_threshold_for_calling.__str__.strip() != '':
90 --standard_min_confidence_threshold_for_calling $analysis_param_type.standard_min_confidence_threshold_for_calling
91 #end if
92 #if $analysis_param_type.standard_min_confidence_threshold_for_emitting.__str__.strip() != '':
93 --standard_min_confidence_threshold_for_emitting $analysis_param_type.standard_min_confidence_threshold_for_emitting
94 #end if
95 #if $analysis_param_type.gcpHMM.__str__.strip() != '':
96 --gcpHMM $analysis_param_type.gcpHMM
97 #end if
98 #if $analysis_param_type.max_alternate_alleles.__str__.strip() != '':
99 --max_alternate_alleles $analysis_param_type.max_alternate_alleles
100 #end if
101 ## mode selections
102 #if $analysis_param_type.genotyping_mode.__str__ != "None" and len($analysis_param_type.genotyping_mode.__str__) > 0:
103 --genotyping_mode $analysis_param_type.genotyping_mode
104 #end if
105 #if $analysis_param_type.pair_hmm_implementation.__str__ != "None" and len($analysis_param_type.pair_hmm_implementation.__str__) > 0:
106 --pair_hmm_implementation $analysis_param_type.pair_hmm_implementation
107 #end if
108 ## optional outputs
109 #if $analysis_param_type.activeRegionOut:
110 --activeRegionOut $active_region_out
111 #end if
112 #if $analysis_param_type.graphOutput:
113 --graphOutput $graph_out
114 #end if
115 ## flags
116 $analysis_param_type.useAllelesTrigger
117 $analysis_param_type.fullHaplotype
118 $analysis_param_type.genotypeFullActiveRegion
119 $analysis_param_type.debug
120 '
121 #end if
122 </command>
123 <inputs>
124 <param name="input_recal" type="data" format="gatk_report" optional="true" label="Covariates table recalibration file" help="-BQSR,--BQSR &amp;lt;recal_file&amp;gt;" >
125 <help>The input covariates table file which enables on-the-fly base quality score recalibration.
126 Enables on-the-fly recalibrate of base qualities. The covariates tables are produced by the BaseQualityScoreRecalibrator tool.
127 Please be aware that one should only run recalibration with the covariates file created on the same input bam(s).
128 </help>
129 </param>
130 <conditional name="reference_source">
131 <expand macro="reference_source_selector_param" />
132 <when value="cached">
133 <param name="input_bam" type="data" format="bam" label="BAM file" help="-I,--input_file &amp;lt;input_file&amp;gt;">
134 <validator type="unspecified_build" />
135 <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 -->
136 </param>
137 <param name="ref_file" type="select" label="Using reference genome" help="-R,--reference_sequence &amp;lt;reference_sequence&amp;gt;" >
138 <options from_data_table="gatk2_picard_indexes">
139 <filter type="data_meta" key="dbkey" ref="input_bam" column="dbkey"/>
140 </options>
141 <validator type="no_options" message="A built-in reference genome is not available for the build associated with the selected input file"/>
142 </param>
143 </when>
144 <when value="history">
145 <param name="input_bam" type="data" format="bam" label="BAM file" help="-I,--input_file &amp;lt;input_file&amp;gt;" />
146 <param name="ref_file" type="data" format="fasta" label="Using reference file" help="-R,--reference_sequence &amp;lt;reference_sequence&amp;gt;">
147 <options>
148 <filter type="data_meta" key="dbkey" ref="input_bam" />
149 </options>
150 </param>
151 </when>
152 </conditional>
153
154 <expand macro="gatk_param_type_conditional" />
155
156 <conditional name="analysis_param_type">
157 <param name="analysis_param_type_selector" type="select" label="Basic or Advanced Analysis options">
158 <option value="basic" selected="True">Basic</option>
159 <option value="advanced">Advanced</option>
160 </param>
161 <when value="basic">
162 <!-- Do nothing here -->
163 </when>
164 <when value="advanced">
165
166 <param name="activeRegionIn" type="data" format="bed,gatk_interval,picard_interval_list,vcf" optional="true" label="activeRegionIn" help="--activeRegionIn / -AR Use this interval list file as the active regions to process"/>
167 <param name="activeRegionOut" type="boolean" checked="False" truevalue="" falsevalue="" label="activeRegionOut" help="--activeRegionOut / -ARO Output the active region to an interval list file"/>
168
169 <param name="annotation" type="select" multiple="True" display="checkboxes" label="Annotation Types" help="-A,--annotation &amp;lt;annotation&amp;gt;">
170 <!-- load the available annotations from an external configuration file, since additional ones can be added to local installs -->
171 <options from_data_table="gatk2_annotations">
172 <filter type="multiple_splitter" column="tools_valid_for" separator=","/>
173 <filter type="static_value" value="UnifiedGenotyper" column="tools_valid_for"/>
174 </options>
175 </param>
176 <repeat name="additional_annotations" title="Additional annotation" help="-A,--annotation &amp;lt;annotation&amp;gt;">
177 <param name="additional_annotation_name" type="text" value="" label="Annotation name" />
178 </repeat>
179 <!--
180 <conditional name="snpEff_rod_bind_type">
181 <param name="snpEff_rod_bind_type_selector" type="select" label="Provide a snpEff reference-ordered data file">
182 <option value="set_snpEff">Set snpEff</option>
183 <option value="exclude_snpEff" selected="True">Don't set snpEff</option>
184 </param>
185 <when value="exclude_snpEff">
186 </when>
187 <when value="set_snpEff">
188 <param name="snpEff_input_rod" type="data" format="vcf" label="ROD file" />
189 <param name="snpEff_rod_name" type="hidden" value="snpEff" label="ROD Name"/>
190 </when>
191 </conditional>
192 -->
193 <param name="group" type="select" multiple="True" display="checkboxes" label="Annotation Interfaces/Groups" help="-G,--group &amp;lt;group&amp;gt;">
194 <option value="RodRequiringAnnotation">RodRequiringAnnotation</option>
195 <option value="Standard">Standard</option>
196 <option value="Experimental">Experimental</option>
197 <option value="WorkInProgress">WorkInProgress</option>
198 <option value="RankSumTest">RankSumTest</option>
199 <!-- <option value="none">none</option> -->
200 </param>
201 <!-- <param name="family_string" type="text" value="" label="Family String"/> -->
202 <param name="exclude_annotations" type="select" multiple="True" display="checkboxes" label="Annotations to exclude" help="-XA,--excludeAnnotation &amp;lt;excludeAnnotation&amp;gt;" >
203 <!-- load the available annotations from an external configuration file, since additional ones can be added to local installs -->
204 <options from_data_table="gatk2_annotations">
205 <filter type="multiple_splitter" column="tools_valid_for" separator=","/>
206 <filter type="static_value" value="UnifiedGenotyper" column="tools_valid_for"/>
207 </options>
208 </param>
209
210 <param name="comp" type="data" format="vcf" optional="true" label="comp" help="--comp / -comp comparison VCF file"/>
211 <param name="contamination_fraction_to_filter" type="float" value="0.05" optional="true" label="contamination_fraction_to_filter" help="--contamination_fraction_to_filter / -contamination Fraction of contamination in sequencing data (for all samples) to aggressively remove">
212 <validator type="in_range" message="value between 0.00 and 1.00" min="0" max="1"/>
213 </param>
214 <param name="dbsnp" type="data" format="vcf" optional="true" label="dbsnp" help="--dbsnp / -D dbSNP file"/>
215 <param name="debug" type="boolean" checked="False" truevalue="-debug" falsevalue="" label="debug" help="--debug / -debug If specified, print out very verbose debug information about each triggering active region"/>
216 <param name="downsampleRegion" type="integer" value="1000" optional="true" label="downsampleRegion" help="--downsampleRegion / -dr coverage, per-sample, to downsample each active region to"/>
217
218 <conditional name="genotyping_mode_type">
219 <param name="genotyping_mode" type="select" label="How to determine the alternate allele to use for genotyping" help="-gt_mode,--genotyping_mode &amp;lt;genotyping_mode&amp;gt;">
220 <option value="DISCOVERY" selected="True">DISCOVERY</option>
221 <option value="GENOTYPE_GIVEN_ALLELES">GENOTYPE_GIVEN_ALLELES</option>
222 </param>
223 <when value="DISCOVERY">
224 <!-- Do nothing here -->
225 </when>
226 <when value="GENOTYPE_GIVEN_ALLELES">
227 <param name="input_alleles_rod" type="data" format="vcf" label="Alleles ROD file" help="-alleles,--alleles &amp;lt;alleles&amp;gt;" />
228 </when>
229 </conditional>
230
231
232 <param name="graphOutput" type="boolean" checked="False" truevalue="" falsevalue="" label="graphOutput" help="--graphOutput / -graph File to which debug assembly graph information should be written"/>
233 <param name="heterozygosity" type="float" value="0.0010" optional="true" label="heterozygosity" help="--heterozygosity / -hets Heterozygosity value used to compute prior likelihoods for any locus"/>
234 <param name="minPruning" type="integer" value="1" optional="true" label="minPruning" help="--minPruning / -minPruning The minimum allowed pruning factor in assembly graph. Paths with &gt;= X supporting kmers are pruned from the graph">
235 <validator type="in_range" message="value between 0 and 127" min="0" max="127"/>
236 </param>
237 <param name="output_mode" type="select" optional="true" label="output_mode" help="--output_mode / -out_mode Specifies which type of calls we should output">
238 <option value="EMIT_VARIANTS_ONLY" selected="True">EMIT_VARIANTS_ONLY</option>
239 <option value="EMIT_ALL_CONFIDENT_SITES">EMIT_ALL_CONFIDENT_SITES</option>
240 <option value="EMIT_ALL_SITES">EMIT_ALL_SITES</option>
241 </param>
242 <param name="pair_hmm_implementation" type="select" optional="true" label="pair_hmm_implementation" help="--pair_hmm_implementation / -pairHMM The PairHMM implementation to use for genotype likelihood calculations">
243 <option value="EXACT">EXACT</option>
244 <option value="ORIGINAL">ORIGINAL</option>
245 <option value="CACHING">CACHING</option>
246 <option value="LOGLESS_CACHING" selected="True">LOGLESS_CACHING</option>
247 </param>
248 <param name="standard_min_confidence_threshold_for_calling" type="float" value="30.0" optional="true" label="standard_min_confidence_threshold_for_calling" help="--standard_min_confidence_threshold_for_calling / -stand_call_conf The minimum phred-scaled confidence threshold at which variants should be called"/>
249 <param name="standard_min_confidence_threshold_for_emitting" type="float" value="30.0" optional="true" label="standard_min_confidence_threshold_for_emitting" help="--standard_min_confidence_threshold_for_emitting / -stand_emit_conf The minimum phred-scaled confidence threshold at which variants should be emitted (and filtered with LowQual if less than the calling threshold)"/>
250 <param name="useAllelesTrigger" type="boolean" checked="False" truevalue="-allelesTrigger" falsevalue="" label="useAllelesTrigger" help="--useAllelesTrigger / -allelesTrigger If specified, use additional trigger on variants found in an external alleles file"/>
251 <param name="fullHaplotype" type="boolean" checked="False" truevalue="-fullHaplotype" falsevalue="" label="fullHaplotype" help="--fullHaplotype / -fullHaplotype If specified, output the full haplotype sequence instead of converting to individual variants w.r.t. the reference"/>
252 <param name="gcpHMM" type="integer" value="10" optional="true" label="gcpHMM" help="--gcpHMM / -gcpHMM Flat gap continuation penalty for use in the Pair HMM"/>
253 <param name="genotypeFullActiveRegion" type="boolean" checked="False" truevalue="-genotypeFullActiveRegion" falsevalue="" label="genotypeFullActiveRegion" help="--genotypeFullActiveRegion / -genotypeFullActiveRegion If specified, alternate alleles are considered to be the full active region for the purposes of genotyping"/>
254 <param name="max_alternate_alleles" type="integer" value="6" optional="true" label="max_alternate_alleles" help="--max_alternate_alleles / -maxAltAlleles Maximum number of alternate alleles to genotype"/>
255 <param name="p_nonref_model" type="select" optional="true" label="p_nonref_model" help="--p_nonref_model / -pnrm Non-reference probability calculation model to employ">
256 <option value="EXACT_INDEPENDENT" selected="True">EXACT_INDEPENDENT experimental implementation - for testing only</option>
257 <option value="EXACT_REFERENCE">EXACT_REFERENCE reference implementation of multi-allelic EXACT model. Extremely slow for many alternate alleles</option>
258 <option value="EXACT_ORIGINAL">EXACT_ORIGINAL original biallelic exact model, for testing only</option>
259 <option value="EXACT_GENERAL_PLOIDY">implementation that supports any sample ploidy</option>
260 </param>
261
262 </when>
263 </conditional>
264 </inputs>
265 <outputs>
266 <data format="vcf" name="output_vcf" label="${tool.name} on ${on_string} (VCF)" />
267 <data format="vcf" name="graph_out" label="${tool.name} on ${on_string} graph" >
268 <filter>analysis_param_type['analysis_param_type_selector'] == "advanced" and analysis_param_type['graphOutput'] == True</filter>
269 </data>
270 <data format="vcf" name="active_region_out" label="${tool.name} on ${on_string} activeRegion" >
271 <filter>analysis_param_type['analysis_param_type_selector'] == "advanced" and analysis_param_type['activeRegionOut'] == True</filter>
272 </data>
273 <data format="txt" name="output_log" label="${tool.name} on ${on_string} (log)" />
274 </outputs>
275 <tests>
276 <test>
277 <param name="input_recal" value="gatk/gatk_count_covariates/gatk_count_covariates_out_1.csv" ftype="csv" />
278 <param name="reference_source_selector" value="history" />
279 <param name="ref_file" value="phiX.fasta" ftype="fasta" />
280 <param name="input_bam" value="gatk/gatk_indel_realigner/gatk_indel_realigner_out_1.bam" ftype="bam" />
281 <param name="gatk_param_type_selector" value="basic" />
282 <param name="analysis_param_type_selector" value="basic" />
283 <output name="output_bam" file="gatk/gatk_table_recalibration/gatk_table_recalibration_out_1.bam" ftype="bam" lines_diff="4" />
284 <output name="output_log" file="gatk/gatk_table_recalibration/gatk_table_recalibration_out_1.log.contains" compare="contains" />
285 </test>
286 </tests>
287 <help>
288 **What it does**
289
290 **HaplotypeCaller**
291 calls SNPs and indels simultaneously via local de-novo assembly of haplotypes in an active region.
292 Haplotypes are evaluated using an affine gap penalty Pair HMM.
293
294 For more information on using read based compression in the GATK, see this `tool specific page &lt;http://www.broadinstitute.org/gatk/gatkdocs/org_broadinstitute_sting_gatk_walkers_haplotypecaller_HaplotypeCaller.html&gt;`_.
295
296 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;`_.
297
298 If you encounter errors, please view the `GATK FAQ &lt;http://www.broadinstitute.org/gatk/guide/topic?name=faqs&gt;`_.
299
300 ------
301
302 **Inputs**
303
304 GenomeAnalysisTK: PrintReads accepts aligned BAM files.
305
306
307 **Outputs**
308
309 The output is a VCF file with raw, unrecalibrated SNP and indel calls.
310
311
312 Go `here &lt;http://www.broadinstitute.org/gatk/guide/topic?name=intro&gt;`_ for details on GATK file formats.
313
314 -------
315
316 **Settings**::
317
318 activeRegionIn Use this interval list file as the active regions to process
319 activeRegionOut Output the active region to this interval list file
320 alleles The set of alleles at which to genotype when --genotyping_mode is GENOTYPE_GIVEN_ALLELES
321 annotation One or more specific annotations to apply to variant calls
322 comp comparison VCF file
323 contamination Fraction of contamination in sequencing data (for all samples) to aggressively remove
324 dbsnp dbSNP file
325 debug If specified, print out very verbose debug information about each triggering active region
326 downsampleRegion coverage, per-sample, to downsample each active region to
327 excludeAnnotation One or more specific annotations to exclude
328 genotyping_mode Specifies how to determine the alternate alleles to use for genotyping
329 graphOutput File to which debug assembly graph information should be written
330 group One or more classes/groups of annotations to apply to variant calls
331 heterozygosity Heterozygosity value used to compute prior likelihoods for any locus
332 minPruning The minimum allowed pruning factor in assembly graph. Paths with less than or equal supporting kmers are pruned from the graph
333 output_mode Specifies which type of calls we should output
334 pair_hmm_implementation The PairHMM implementation to use for genotype likelihood calculations
335 stand_call_conf The minimum phred-scaled confidence threshold at which variants should be called
336 stand_emit_conf The minimum phred-scaled confidence threshold at which variants should be emitted (and filtered with LowQual if less than the calling threshold)
337 useAllelesTrigger If specified, use additional trigger on variants found in an external alleles file
338 fullHaplotype If specified, output the full haplotype sequence instead of converting to individual variants w.r.t. the reference
339 gcpHMM Flat gap continuation penalty for use in the Pair HMM
340 genotypeFullActiveRegion If specified, alternate alleles are considered to be the full active region for the purposes of genotyping
341 max_alternate_alleles Maximum number of alternate alleles to genotype
342 p_nonref_model Non-reference probability calculation model to employ
343
344 ------
345
346 @CITATION_SECTION@
347 </help>
348 </tool>