|
0
|
1 <tool id="gatk_variant_eval" name="Eval Variants" version="0.0.8">
|
|
|
2 <description></description>
|
|
|
3 <requirements>
|
|
|
4 <requirement type="package" version="1.4">gatk</requirement>
|
|
|
5 </requirements>
|
|
|
6 <macros>
|
|
|
7 <import>gatk_macros.xml</import>
|
|
|
8 </macros>
|
|
|
9 <command interpreter="python">gatk_wrapper.py
|
|
|
10 #from binascii import hexlify
|
|
|
11 --max_jvm_heap_fraction "1"
|
|
|
12 --stdout "${output_log}"
|
|
|
13 #for $var_count, $variant in enumerate( $reference_source.variants ):
|
|
|
14 -d "--eval:input_${var_count},%(file_type)s" "${variant.input_variant}" "${variant.input_variant.ext}" "input_variants_${var_count}"
|
|
|
15 #end for
|
|
|
16 -p 'java
|
|
|
17 -jar "\$JAVA_JAR_PATH/GenomeAnalysisTK.jar"
|
|
|
18 -T "VariantEval"
|
|
|
19 --out "${output_report}"
|
|
|
20 --num_threads \${GALAXY_SLOTS:-4}
|
|
|
21 -et "NO_ET" ##ET no phone home
|
|
|
22 ##-log "${output_log}" ##don't use this to log to file, instead directly capture stdout
|
|
|
23 #if $reference_source.reference_source_selector != "history":
|
|
|
24 -R "${reference_source.ref_file.fields.path}"
|
|
|
25 #end if
|
|
|
26 '
|
|
|
27
|
|
|
28 #for $rod_binding in $comp_rod_bind:
|
|
|
29 -d "--comp:${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}"
|
|
|
30 #if str( $rod_binding.comp_known_names ):
|
|
|
31 -p '--known_names "${rod_binding.comp_rod_name}"'
|
|
|
32 #end if
|
|
|
33 #end for
|
|
|
34
|
|
|
35 #if str( $dbsnp_rod_bind_type.dbsnp_rod_bind_type_selector ) == 'set_dbsnp':
|
|
|
36 -d "--dbsnp:${dbsnp_rod_bind_type.dbsnp_rod_name},%(file_type)s" "${dbsnp_rod_bind_type.dbsnp_input_rod}" "${dbsnp_rod_bind_type.dbsnp_input_rod.ext}" "input_dbsnp_${dbsnp_rod_bind_type.dbsnp_rod_name}"
|
|
|
37 #if str( $dbsnp_rod_bind_type.dbsnp_known_names ):
|
|
|
38 -p '--known_names "${dbsnp_rod_bind_type.dbsnp_rod_name}"'
|
|
|
39 #end if
|
|
|
40 #end if
|
|
|
41
|
|
|
42 #include source=$standard_gatk_options#
|
|
|
43
|
|
|
44
|
|
|
45 ##start analysis specific options
|
|
|
46 #if $analysis_param_type.analysis_param_type_selector == "advanced":
|
|
|
47 #for $stratification in $analysis_param_type.stratifications:
|
|
|
48 #set $select_string = "--select_exps '%s' --select_names '%s'" % ( str( $stratification.select_exps ), str( $stratification.select_name ) )
|
|
|
49 -o '${ hexlify( $select_string ) }'
|
|
|
50 #end for
|
|
|
51 -p '
|
|
|
52
|
|
|
53 #for $sample in $analysis_param_type.samples:
|
|
|
54 --sample "${sample.sample}"
|
|
|
55 #end for
|
|
|
56
|
|
|
57 #if str( $analysis_param_type.stratification_modules ) != "None":
|
|
|
58 #for $stratification_module in str( $analysis_param_type.stratification_modules).split( ',' ):
|
|
|
59 --stratificationModule "${stratification_module}"
|
|
|
60 #end for
|
|
|
61 #end if
|
|
|
62
|
|
|
63 ${analysis_param_type.do_not_use_all_standard_stratifications}
|
|
|
64
|
|
|
65 #for $variant_type in $analysis_param_type.only_variants_of_type:
|
|
|
66 --onlyVariantsOfType "${variant_type.variant_type}"
|
|
|
67 #end for
|
|
|
68
|
|
|
69 #if str( $analysis_param_type.eval_modules ) != "None":
|
|
|
70 #for $eval_module in str( $analysis_param_type.eval_modules).split( ',' ):
|
|
|
71 --evalModule "${eval_module}"
|
|
|
72 #end for
|
|
|
73 #end if
|
|
|
74
|
|
|
75 ${analysis_param_type.do_not_use_all_standard_modules}
|
|
|
76
|
|
|
77 #if str( $analysis_param_type.num_samples ) != "0":
|
|
|
78 --numSamples "${analysis_param_type.num_samples}"
|
|
|
79 #end if
|
|
|
80
|
|
|
81 --minPhaseQuality "${analysis_param_type.min_phase_quality}"
|
|
|
82
|
|
|
83 #if str( $analysis_param_type.family ):
|
|
|
84 --family_structure "${analysis_param_type.family}"
|
|
|
85 #end if
|
|
|
86
|
|
|
87 --mendelianViolationQualThreshold "${analysis_param_type.mendelian_violation_qual_threshold}"
|
|
|
88
|
|
|
89 #if str( $analysis_param_type.ancestral_alignments ) != "None":
|
|
|
90 --ancestralAlignments "${analysis_param_type.ancestral_alignments}"
|
|
|
91 #end if
|
|
|
92 '
|
|
|
93 #if str( $analysis_param_type.known_cnvs ) != "None":
|
|
|
94 -d "--knownCNVs" "${analysis_param_type.known_cnvs}" "${analysis_param_type.known_cnvs.ext}" "input_known_cnvs"
|
|
|
95 #end if
|
|
|
96
|
|
|
97 #if str( $analysis_param_type.strat_intervals ) != "None":
|
|
|
98 -d "--stratIntervals" "${analysis_param_type.strat_intervals}" "${analysis_param_type.strat_intervals.ext}" "input_strat_intervals"
|
|
|
99 #end if
|
|
|
100 #end if
|
|
|
101 </command>
|
|
|
102 <inputs>
|
|
|
103
|
|
|
104 <conditional name="reference_source">
|
|
|
105 <expand macro="reference_source_selector_param" />
|
|
|
106 <when value="cached">
|
|
|
107 <repeat name="variants" title="Variant" min="1" help="-eval,--eval &lt;eval&gt;">
|
|
|
108 <param name="input_variant" type="data" format="vcf" label="Input variant file" />
|
|
|
109 </repeat>
|
|
|
110 <param name="ref_file" type="select" label="Using reference genome" help="-R,--reference_sequence &lt;reference_sequence&gt;">
|
|
|
111 <options from_data_table="gatk_picard_indexes">
|
|
|
112 <!-- <filter type="data_meta" key="dbkey" ref="input_variant" column="dbkey"/> -->
|
|
|
113 </options>
|
|
|
114 <validator type="no_options" message="A built-in reference genome is not available for the build associated with the selected input file"/>
|
|
|
115 </param>
|
|
|
116 </when>
|
|
|
117 <when value="history"> <!-- FIX ME!!!! -->
|
|
|
118 <repeat name="variants" title="Variant" min="1" help="-eval,--eval &lt;eval&gt;">
|
|
|
119 <param name="input_variant" type="data" format="vcf" label="Input variant file" />
|
|
|
120 </repeat>
|
|
|
121 <param name="ref_file" type="data" format="fasta" label="Using reference file" help="-R,--reference_sequence &lt;reference_sequence&gt;" />
|
|
|
122 </when>
|
|
|
123 </conditional>
|
|
|
124
|
|
|
125 <repeat name="comp_rod_bind" title="Binding for reference-ordered comparison data" help="-comp,--comp &lt;comp&gt;">
|
|
|
126 <param name="comp_input_rod" type="data" format="vcf" label="Comparison ROD file" />
|
|
|
127 <param name="comp_rod_name" type="text" value="Unnamed" label="Comparison ROD Name"/>
|
|
|
128 <param name="comp_known_names" type="boolean" truevalue="--known_names" falsevalue="" label="Use Comparison ROD as known_names" help="-knownName,--known_names &lt;known_names&gt;"/>
|
|
|
129 </repeat>
|
|
|
130
|
|
|
131 <conditional name="dbsnp_rod_bind_type">
|
|
|
132 <param name="dbsnp_rod_bind_type_selector" type="select" label="Provide a dbSNP reference-ordered data file" help="-D,--dbsnp &lt;dbsnp&gt;">
|
|
|
133 <option value="set_dbsnp" selected="True">Set dbSNP</option>
|
|
|
134 <option value="exclude_dbsnp">Don't set dbSNP</option>
|
|
|
135 </param>
|
|
|
136 <when value="exclude_dbsnp">
|
|
|
137 <!-- Do nothing here -->
|
|
|
138 </when>
|
|
|
139 <when value="set_dbsnp">
|
|
|
140 <param name="dbsnp_input_rod" type="data" format="vcf" label="dbSNP ROD file" />
|
|
|
141 <param name="dbsnp_rod_name" type="hidden" value="dbsnp" label="dbSNP ROD Name"/>
|
|
|
142 <param name="dbsnp_known_names" type="boolean" truevalue="--known_names" falsevalue="" label="Use dbSNP ROD as known_names" help="-knownName,--known_names &lt;known_names&gt;" />
|
|
|
143 </when>
|
|
|
144 </conditional>
|
|
|
145
|
|
|
146 <expand macro="gatk_param_type_conditional" />
|
|
|
147
|
|
|
148
|
|
|
149 <expand macro="analysis_type_conditional">
|
|
|
150 <repeat name="stratifications" title="Stratification">
|
|
|
151 <param name="select_exps" value="" type="text" label="Stratification Expression" help="-select,--select_exps &lt;select_exps&gt;">
|
|
|
152 <sanitizer>
|
|
|
153 <valid initial="string.printable">
|
|
|
154 <remove value="'"/>
|
|
|
155 </valid>
|
|
|
156 <mapping initial="none"/>
|
|
|
157 </sanitizer>
|
|
|
158 </param>
|
|
|
159 <param name="select_name" value="" type="text" label="Name" help="-selectName,--select_names &lt;select_names&gt;"/>
|
|
|
160 </repeat>
|
|
|
161
|
|
|
162 <repeat name="samples" title="Sample" help="-sn,--sample &lt;sample&gt;">
|
|
|
163 <param name="sample" value="" type="text" label="Derive eval and comp contexts using only these sample genotypes, when genotypes are available in the original context"/>
|
|
|
164 </repeat>
|
|
|
165
|
|
|
166 <param name="stratification_modules" type="select" multiple="True" display="checkboxes" label="Stratification modules to apply to the eval track(s)" help="-ST,--stratificationModule &lt;stratificationModule&gt;" >
|
|
|
167 <!-- do these need individual options also? gatk wiki has little info -->
|
|
|
168 <option value="AlleleFrequency" />
|
|
|
169 <option value="AlleleCount" />
|
|
|
170 <option value="CompRod" />
|
|
|
171 <option value="Contig" />
|
|
|
172 <option value="CpG" />
|
|
|
173 <option value="Degeneracy" />
|
|
|
174 <option value="EvalRod" />
|
|
|
175 <option value="Filter" />
|
|
|
176 <option value="FunctionalClass" />
|
|
|
177 <option value="JexlExpression" />
|
|
|
178 <option value="Sample" />
|
|
|
179 <option value="IntervalStratification" />
|
|
|
180 </param>
|
|
|
181 <param name="do_not_use_all_standard_stratifications" checked="false" type="boolean" truevalue="--doNotUseAllStandardStratifications" falsevalue="" label="Do not use the standard stratification modules by default" help="-noST,--doNotUseAllStandardStratifications" />
|
|
|
182
|
|
|
183 <repeat name="only_variants_of_type" title="only Variants Of Type" help="--onlyVariantsOfType">
|
|
|
184 <param name="variant_type" type="text" value="" label="only variants of these types will be considered during the evaluation"/>
|
|
|
185 </repeat>
|
|
|
186
|
|
|
187 <param name="eval_modules" type="select" multiple="True" display="checkboxes" label="Eval modules to apply to the eval track(s)" help="-EV,--evalModule &lt;evalModule&gt;" >
|
|
|
188 <!-- do these need individual options also? gatk wiki has little info -->
|
|
|
189 <option value="ACTransitionTable" />
|
|
|
190 <option value="AlleleFrequencyComparison" />
|
|
|
191 <option value="AminoAcidTransition" />
|
|
|
192 <option value="CompOverlap" />
|
|
|
193 <option value="CountVariants" />
|
|
|
194 <option value="GenotypeConcordance" />
|
|
|
195 <option value="GenotypePhasingEvaluator" />
|
|
|
196 <option value="IndelMetricsByAC" />
|
|
|
197 <option value="IndelStatistics" />
|
|
|
198 <option value="MendelianViolationEvaluator" />
|
|
|
199 <option value="PrintMissingComp" />
|
|
|
200 <option value="PrivatePermutations" />
|
|
|
201 <option value="SimpleMetricsByAC" />
|
|
|
202 <option value="ThetaVariantEvaluator" />
|
|
|
203 <option value="TiTvVariantEvaluator" />
|
|
|
204 <option value="VariantQualityScore" />
|
|
|
205 </param>
|
|
|
206 <param name="do_not_use_all_standard_modules" checked="false" type="boolean" truevalue="--doNotUseAllStandardModules" falsevalue="" label="Do not use the standard eval modules by default" help="-noEV,--doNotUseAllStandardModules" />
|
|
|
207
|
|
|
208 <param name="num_samples" type="integer" label="Number of samples (used if no samples are available in the VCF file" value="0" help="-ns,--numSamples &lt;numSamples&gt;"/>
|
|
|
209 <param name="min_phase_quality" type="float" label="Minimum phasing quality " value="10.0" help="-mpq,--minPhaseQuality &lt;minPhaseQuality&gt;"/>
|
|
|
210 <param name="family" type="text" value="" label="If provided, genotypes in will be examined for mendelian violations: this argument is a string formatted as dad+mom=child where these parameters determine which sample names are examined" help="--family_structure"/>
|
|
|
211 <param name="mendelian_violation_qual_threshold" type="integer" label="Minimum genotype QUAL score for each trio member required to accept a site as a violation" value="50" help="-mvq,--mendelianViolationQualThreshold &lt;mendelianViolationQualThreshold&gt;"/>
|
|
|
212 <param name="ancestral_alignments" type="data" format="fasta" optional="True" label="Fasta file with ancestral alleles" help="-aa,--ancestralAlignments &lt;ancestralAlignments&gt;" />
|
|
|
213 <param name="known_cnvs" type="data" format="bed,gatk_interval,picard_interval_list" optional="True" label="File containing tribble-readable features describing a known list of copy number variants" help="-knownCNVs,--knownCNVs &lt;knownCNVs&gt;" />
|
|
|
214 <param name="strat_intervals" type="data" format="bed,gatk_interval,picard_interval_list" optional="True" label="File containing tribble-readable features for the IntervalStratificiation" help="-stratIntervals,--stratIntervals &lt;stratIntervals&gt;" />
|
|
|
215
|
|
|
216 </expand>
|
|
|
217
|
|
|
218
|
|
|
219 </inputs>
|
|
|
220 <outputs>
|
|
|
221 <data format="gatk_report" name="output_report" label="${tool.name} on ${on_string} (report)" />
|
|
|
222 <data format="txt" name="output_log" label="${tool.name} on ${on_string} (log)" />
|
|
|
223 </outputs>
|
|
|
224 <tests>
|
|
|
225 <test>
|
|
|
226 <param name="reference_source_selector" value="history" />
|
|
|
227 <param name="ref_file" value="phiX.fasta" ftype="fasta" />
|
|
|
228 <param name="input_variant" value="gatk/gatk_variant_annotator/gatk_variant_annotator_out_1.vcf" ftype="vcf" />
|
|
|
229 <param name="dbsnp_rod_bind_type_selector" value="set_dbsnp" />
|
|
|
230 <param name="dbsnp_input_rod" value="gatk/fake_phiX_variant_locations.vcf" ftype="vcf" />
|
|
|
231 <param name="dbsnp_known_names" value="True"/>
|
|
|
232 <param name="comp_rod_bind" value="0" />
|
|
|
233 <param name="gatk_param_type_selector" value="basic" />
|
|
|
234 <param name="analysis_param_type_selector" value="basic" />
|
|
|
235 <output name="output_report" file="gatk/gatk_variant_eval/gatk_variant_eval_out_1.gatk_report" />
|
|
|
236 <output name="output_log" file="gatk/gatk_variant_eval/gatk_variant_eval_out_1.log.contains" compare="contains" />
|
|
|
237 </test>
|
|
|
238 </tests>
|
|
|
239 <help>
|
|
|
240 **What it does**
|
|
|
241
|
|
|
242 General-purpose tool for variant evaluation (% in dbSNP, genotype concordance, Ti/Tv ratios, and a lot more)
|
|
|
243
|
|
|
244 For more information on using the VariantEval module, see this `tool specific page <http://www.broadinstitute.org/gsa/wiki/index.php/VariantEval>`_.
|
|
|
245
|
|
|
246 To learn about best practices for variant detection using GATK, see this `overview <http://www.broadinstitute.org/gsa/wiki/index.php/Best_Practice_Variant_Detection_with_the_GATK_v3>`_.
|
|
|
247
|
|
|
248 If you encounter errors, please view the `GATK FAQ <http://www.broadinstitute.org/gsa/wiki/index.php/Frequently_Asked_Questions>`_.
|
|
|
249
|
|
|
250 ------
|
|
|
251
|
|
|
252 **Inputs**
|
|
|
253
|
|
|
254 GenomeAnalysisTK: VariantEval accepts variant files as input.
|
|
|
255
|
|
|
256
|
|
|
257 **Outputs**
|
|
|
258
|
|
|
259 The output is a table of variant evaluation.
|
|
|
260
|
|
|
261
|
|
|
262 Go `here <http://www.broadinstitute.org/gsa/wiki/index.php/Input_files_for_the_GATK>`_ for details on GATK file formats.
|
|
|
263
|
|
|
264
|
|
|
265 -------
|
|
|
266
|
|
|
267 **Settings**::
|
|
|
268
|
|
|
269 out An output file presented to the walker. Will overwrite contents if file exists.
|
|
|
270 list List the available eval modules and exit
|
|
|
271 select_exps One or more stratifications to use when evaluating the data
|
|
|
272 select_names Names to use for the list of stratifications (must be a 1-to-1 mapping)
|
|
|
273 sample Derive eval and comp contexts using only these sample genotypes, when genotypes are available in the original context
|
|
|
274 known_names Name of ROD bindings containing variant sites that should be treated as known when splitting eval rods into known and novel subsets
|
|
|
275 stratificationModule One or more specific stratification modules to apply to the eval track(s) (in addition to the standard stratifications, unless -noS is specified)
|
|
|
276 doNotUseAllStandardStratifications Do not use the standard stratification modules by default (instead, only those that are specified with the -S option)
|
|
|
277 onlyVariantsOfType If provided, only variants of these types will be considered during the evaluation, in
|
|
|
278 evalModule One or more specific eval modules to apply to the eval track(s) (in addition to the standard modules, unless -noE is specified)
|
|
|
279 doNotUseAllStandardModules Do not use the standard modules by default (instead, only those that are specified with the -E option)
|
|
|
280 numSamples Number of samples (used if no samples are available in the VCF file
|
|
|
281 minPhaseQuality Minimum phasing quality
|
|
|
282 family_structure If provided, genotypes in will be examined for mendelian violations: this argument is a string formatted as dad+mom=child where these parameters determine which sample names are examined
|
|
|
283 mendelianViolationQualThreshold Minimum genotype QUAL score for each trio member required to accept a site as a violation
|
|
|
284 ancestralAlignments Fasta file with ancestral alleles
|
|
|
285
|
|
|
286 @CITATION_SECTION@
|
|
|
287 </help>
|
|
|
288 </tool>
|
