Mercurial > repos > lz_hust > gatktools
comparison variant_eval.xml @ 15:01ff8dd37d4d draft default tip
Uploaded
author | lz_hust |
---|---|
date | Sat, 01 Jun 2019 07:20:41 -0400 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
14:68426930d59c | 15:01ff8dd37d4d |
---|---|
1 <tool id="gatk2_variant_eval" name="Eval Variants" version="@VERSION@.1"> | |
2 <description></description> | |
3 <macros> | |
4 <import>gatk2_macros.xml</import> | |
5 </macros> | |
6 <expand macro="requirements" /> | |
7 <expand macro="version_command" /> | |
8 <command interpreter="python"> | |
9 #from binascii import hexlify | |
10 | |
11 gatk2_wrapper.py | |
12 --stdout "${output_log}" | |
13 #for $var_count, $variant in enumerate( $reference_source.input_variants ): | |
14 -d "--eval:input_${var_count},%(file_type)s" "${variant}" "${variant.ext}" "input_variants_${var_count}" | |
15 #end for | |
16 -p ' | |
17 @JAR_PATH@ | |
18 -T "VariantEval" | |
19 --out "${output_report}" | |
20 \$GATK2_SITE_OPTIONS | |
21 | |
22 @THREADS@ | |
23 | |
24 #if $reference_source.reference_source_selector != "history": | |
25 -R "${reference_source.ref_file.fields.path}" | |
26 #end if | |
27 ' | |
28 | |
29 #for $rod_binding in $comp_rod_bind: | |
30 -d "--comp:${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}" | |
31 #if str( $rod_binding.comp_known_names ): | |
32 -p '--known_names "${rod_binding.comp_rod_name}"' | |
33 #end if | |
34 #end for | |
35 | |
36 #if $dbsnp_rod_bind_type.dbsnp_rod_bind_type_selector == 'set_dbsnp' | |
37 -d "--dbsnp:${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}" | |
38 #if $dbsnp_rod_bind_type.dbsnp_known_names | |
39 -p '--known_names "${dbsnp_rod_bind_type.dbsnp_rod_name}"' | |
40 #end if | |
41 #end if | |
42 | |
43 #include source=$standard_gatk_options# | |
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 "${analysis_param_type.min_phase_quality}" | |
82 | |
83 --mendelianViolationQualThreshold "${analysis_param_type.mendelian_violation_qual_threshold}" | |
84 | |
85 #if str( $analysis_param_type.ancestral_alignments ) != "None": | |
86 --ancestralAlignments "${analysis_param_type.ancestral_alignments}" | |
87 #end if | |
88 ' | |
89 #if str( $analysis_param_type.known_cnvs ) != "None": | |
90 -d "--knownCNVs" "${analysis_param_type.known_cnvs}" "${analysis_param_type.known_cnvs.ext}" "input_known_cnvs" | |
91 #end if | |
92 | |
93 #if str( $analysis_param_type.strat_intervals ) != "None": | |
94 -d "--stratIntervals" "${analysis_param_type.strat_intervals}" "${analysis_param_type.strat_intervals.ext}" "input_strat_intervals" | |
95 #end if | |
96 #end if | |
97 </command> | |
98 <inputs> | |
99 | |
100 <conditional name="reference_source"> | |
101 <expand macro="reference_source_selector_param" /> | |
102 <when value="cached"> | |
103 <expand macro="input_variants" help="-eval,--eval &lt;eval&gt;"/> | |
104 <param name="ref_file" type="select" label="Using reference genome" help="-R,--reference_sequence &lt;reference_sequence&gt;"> | |
105 <options from_data_table="gatk2_picard_indexes"> | |
106 <!-- <filter type="data_meta" key="dbkey" ref="input_variant" column="dbkey"/> --> | |
107 </options> | |
108 <validator type="no_options" message="A built-in reference genome is not available for the build associated with the selected input file"/> | |
109 </param> | |
110 </when> | |
111 <when value="history"> <!-- FIX ME!!!! --> | |
112 <expand macro="input_variants" help="-eval,--eval &lt;eval&gt;" /> | |
113 <param name="ref_file" type="data" format="fasta" label="Using reference file" help="-R,--reference_sequence &lt;reference_sequence&gt;" /> | |
114 </when> | |
115 </conditional> | |
116 | |
117 <repeat name="comp_rod_bind" title="Comparison Reference-Ordered Data (ROD) file" help="-comp,--comp &lt;comp&gt;"> | |
118 <param name="comp_input_rod" type="data" format="vcf" label="Comparison ROD file" /> | |
119 <param name="comp_rod_name" type="text" value="" label="Comparison ROD name"> | |
120 <validator type="regex" message="Value must be a not empty string composed by alphanumeric characters and underscores">^\w+$</validator> | |
121 </param> | |
122 <param name="comp_known_names" type="boolean" label="Use comparison ROD file as known_names" help="-knownName,--known_names &lt;known_names&gt;"/> | |
123 </repeat> | |
124 <conditional name="dbsnp_rod_bind_type"> | |
125 <param name="dbsnp_rod_bind_type_selector" type="select" label="Provide a dbSNP Reference-Ordered Data (ROD) file" help="-D,--dbsnp &lt;dbsnp&gt;"> | |
126 <option value="set_dbsnp" selected="True">Set dbSNP</option> | |
127 <option value="exclude_dbsnp">Don't set dbSNP</option> | |
128 </param> | |
129 <when value="exclude_dbsnp" /> | |
130 <when value="set_dbsnp"> | |
131 <param name="dbsnp_input_rod" type="data" format="vcf" label="dbSNP ROD file" /> | |
132 <param name="dbsnp_rod_name" type="text" value="dbsnp" label="dbsnp ROD name"> | |
133 <validator type="regex" message="Value must be a not empty string composed by alphanumeric characters and underscores">^\w+$</validator> | |
134 </param> | |
135 <param name="dbsnp_known_names" type="boolean" label="Use dbSNP ROD file as known_names" help="-knownName,--known_names &lt;known_names&gt;" /> | |
136 </when> | |
137 </conditional> | |
138 | |
139 <expand macro="gatk_param_type_conditional" /> | |
140 | |
141 <expand macro="analysis_type_conditional"> | |
142 <repeat name="stratifications" title="Stratification"> | |
143 <param name="select_exps" value="" type="text" label="Stratification Expression" help="-select,--select_exps &lt;select_exps&gt;"> | |
144 <sanitizer> | |
145 <valid initial="string.printable"> | |
146 <remove value="'"/> | |
147 </valid> | |
148 <mapping initial="none"/> | |
149 </sanitizer> | |
150 </param> | |
151 <param name="select_name" value="" type="text" label="Name" help="-selectName,--select_names &lt;select_names&gt;"/> | |
152 </repeat> | |
153 | |
154 <repeat name="samples" title="Sample" help="-sn,--sample &lt;sample&gt;"> | |
155 <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"/> | |
156 </repeat> | |
157 | |
158 <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;" > | |
159 <option value="AlleleCount" /> | |
160 <option value="AlleleFrequency" /> | |
161 <option value="CompRod" /> | |
162 <option value="Contig" /> | |
163 <option value="CpG" /> | |
164 <option value="Degeneracy" /> | |
165 <option value="EvalRod" /> | |
166 <option value="Filter" /> | |
167 <option value="FunctionalClass" /> | |
168 <option value="IndelSize" /> | |
169 <option value="IntervalStratification" /> | |
170 <option value="JexlExpression" /> | |
171 <option value="Novelty" /> | |
172 <option value="OneBPIndel" /> | |
173 <option value="Sample" /> | |
174 <option value="SnpEffPositionModifier" /> | |
175 <option value="TandemRepeat" /> | |
176 <option value="VariantType" /> | |
177 </param> | |
178 <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" /> | |
179 | |
180 <repeat name="only_variants_of_type" title="only Variants Of Type" help="--onlyVariantsOfType"> | |
181 <param name="variant_type" type="text" value="" label="only variants of these types will be considered during the evaluation"/> | |
182 </repeat> | |
183 | |
184 <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;" > | |
185 <option value="CompOverlap" /> | |
186 <option value="CountVariants" /> | |
187 <option value="IndelLengthHistogram" /> | |
188 <option value="IndelSummary" /> | |
189 <option value="MendelianViolationEvaluator" /> | |
190 <option value="MultiallelicSummary" /> | |
191 <option value="PrintMissingComp" /> | |
192 <option value="ThetaVariantEvaluator" /> | |
193 <option value="TiTvVariantEvaluator" /> | |
194 <option value="ValidationReport" /> | |
195 <option value="VariantSummary" /> | |
196 </param> | |
197 <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" /> | |
198 | |
199 <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;"/> | |
200 <param name="min_phase_quality" type="float" label="Minimum phasing quality " value="10.0" help="-mpq,--minPhaseQuality &lt;minPhaseQuality&gt;"/> | |
201 <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;"/> | |
202 <param name="ancestral_alignments" type="data" format="fasta" optional="True" label="Fasta file with ancestral alleles" help="-aa,--ancestralAlignments &lt;ancestralAlignments&gt;" /> | |
203 <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;" /> | |
204 <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;" /> | |
205 </expand> | |
206 | |
207 </inputs> | |
208 <outputs> | |
209 <data format="gatk_report" name="output_report" label="${tool.name} on ${on_string} (report)" /> | |
210 <data format="txt" name="output_log" label="${tool.name} on ${on_string} (log)" /> | |
211 </outputs> | |
212 <tests> | |
213 <test> | |
214 <param name="reference_source_selector" value="history" /> | |
215 <param name="ref_file" value="phiX.fasta" ftype="fasta" /> | |
216 <param name="input_variant" value="gatk/gatk_variant_annotator/gatk_variant_annotator_out_1.vcf" ftype="vcf" /> | |
217 <param name="dbsnp_rod_bind_type_selector" value="set_dbsnp" /> | |
218 <param name="dbsnp_input_rod" value="gatk/fake_phiX_variant_locations.vcf" ftype="vcf" /> | |
219 <param name="dbsnp_rod_name" value="dbsnp" /> | |
220 <param name="dbsnp_known_names" value="True"/> | |
221 <param name="comp_rod_bind" value="0" /> | |
222 <param name="gatk_param_type_selector" value="basic" /> | |
223 <param name="analysis_param_type_selector" value="basic" /> | |
224 <output name="output_report" file="gatk/gatk_variant_eval/gatk_variant_eval_out_1.gatk_report" /> | |
225 <output name="output_log" file="gatk/gatk_variant_eval/gatk_variant_eval_out_1.log.contains" compare="contains" /> | |
226 </test> | |
227 </tests> | |
228 <help> | |
229 **What it does** | |
230 | |
231 General-purpose tool for variant evaluation (% in dbSNP, genotype concordance, Ti/Tv ratios, and a lot more) | |
232 | |
233 For more information on using the VariantEval module, see this `tool specific page <http://www.broadinstitute.org/gatk/gatkdocs/org_broadinstitute_sting_gatk_walkers_varianteval_VariantEval.html>`_. | |
234 | |
235 To learn about best practices for variant detection using GATK, see this `overview <http://www.broadinstitute.org/gatk/guide/topic?name=best-practices>`_. | |
236 | |
237 If you encounter errors, please view the `GATK FAQ <http://www.broadinstitute.org/gatk/guide/topic?name=faqs>`_. | |
238 | |
239 ------ | |
240 | |
241 **Inputs** | |
242 | |
243 GenomeAnalysisTK: VariantEval accepts variant files as input. | |
244 | |
245 | |
246 **Outputs** | |
247 | |
248 The output is a table of variant evaluation. | |
249 | |
250 | |
251 Go `here <http://www.broadinstitute.org/gatk/guide/topic?name=intro>`_ for details on GATK file formats. | |
252 | |
253 ------- | |
254 | |
255 **Settings**:: | |
256 | |
257 out An output file presented to the walker. Will overwrite contents if file exists. | |
258 list List the available eval modules and exit | |
259 select_exps One or more stratifications to use when evaluating the data | |
260 select_names Names to use for the list of stratifications (must be a 1-to-1 mapping) | |
261 sample Derive eval and comp contexts using only these sample genotypes, when genotypes are available in the original context | |
262 known_names Name of ROD bindings containing variant sites that should be treated as known when splitting eval rods into known and novel subsets | |
263 stratificationModule One or more specific stratification modules to apply to the eval track(s) (in addition to the standard stratifications, unless -noS is specified) | |
264 doNotUseAllStandardStratifications Do not use the standard stratification modules by default (instead, only those that are specified with the -S option) | |
265 onlyVariantsOfType If provided, only variants of these types will be considered during the evaluation, in | |
266 evalModule One or more specific eval modules to apply to the eval track(s) (in addition to the standard modules, unless -noE is specified) | |
267 doNotUseAllStandardModules Do not use the standard modules by default (instead, only those that are specified with the -E option) | |
268 numSamples Number of samples (used if no samples are available in the VCF file | |
269 minPhaseQuality Minimum phasing quality | |
270 mendelianViolationQualThreshold Minimum genotype QUAL score for each trio member required to accept a site as a violation | |
271 ancestralAlignments Fasta file with ancestral alleles | |
272 | |
273 @CITATION_SECTION@ | |
274 </help> | |
275 <expand macro="citations" /> | |
276 </tool> |