Mercurial > repos > iuc > bcftools_norm

comparison bcftools_norm.xml @ 9:35cf715a87e5 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/bcftools commit e648d86d550ddf2eb67237752320c390b3a780e5
author iuc
date Wed, 05 Jun 2019 12:40:36 -0400
parents 05afd752df18
children 9ef2d8eb8c30
comparison
equal deleted inserted replaced
8:05afd752df18 9:35cf715a87e5
1 <?xml version='1.0' encoding='utf-8'?> 1 <?xml version='1.0' encoding='utf-8'?>
2 <tool name="bcftools @EXECUTABLE@" id="bcftools_@EXECUTABLE@" version="@TOOL_VERSION@"> 2 <tool name="bcftools @EXECUTABLE@" id="bcftools_@EXECUTABLE@" version="@TOOL_VERSION@+galaxy1">
3 <description>Left-align and normalize indels; check if REF alleles match the reference; split multiallelic sites into multiple rows; recover multiallelics from multiple rows</description> 3 <description>Left-align and normalize indels; check if REF alleles match the reference; split multiallelic sites into multiple rows; recover multiallelics from multiple rows</description>
4 <macros> 4 <macros>
5 <token name="@EXECUTABLE@">norm</token> 5 <token name="@EXECUTABLE@">norm</token>
6 <import>macros.xml</import> 6 <import>macros.xml</import>
7 </macros> 7 </macros>
10 </expand> 10 </expand>
11 <expand macro="version_command" /> 11 <expand macro="version_command" />
12 <command detect_errors="aggressive"><![CDATA[ 12 <command detect_errors="aggressive"><![CDATA[
13 @PREPARE_ENV@ 13 @PREPARE_ENV@
14 @PREPARE_INPUT_FILE@ 14 @PREPARE_INPUT_FILE@
15 #set $section = $sec_ref 15 #set $section = $reference_source
16 @PREPARE_FASTA_REF@ 16 @PREPARE_FASTA_REF@
17 #set $section = $sec_restrict 17 #set $section = $sec_restrict
18 @PREPARE_TARGETS_FILE@ 18 @PREPARE_TARGETS_FILE@
19 @PREPARE_REGIONS_FILE@ 19 @PREPARE_REGIONS_FILE@
20 20
21 bcftools @EXECUTABLE@ 21 bcftools @EXECUTABLE@
22 22
23 #set $section = $sec_ref 23 #set $section = $reference_source
24 @FASTA_REF@ 24 @FASTA_REF@
25 ${section.do_not_normalize} 25 --check-ref $check_ref
26 #if $section.check_ref: 26 $normalize_indels
27 --check-ref "${section.check_ref}" 27 #if $rm_dup:
28 #end if 28 --rm-dup "$rm_dup"
29 29 #end if
30 ## Default section 30 #if $multiallelics.mode:
31 #set $section = $sec_default 31 --multiallelics '${multiallelics.mode}${multiallelics.multiallelic_types}'
32 #if $section.rm_dup: 32 #end if
33 --rm-dup "${section.rm_dup}" 33 #if $multiallelics.mode == '+':
34 #end if 34 ${multiallelics.strict_filter}
35 #if $section.multiallelics: 35 #end if
36 #set $mtype = '' 36 #if $sec_default.site_win:
37 #if $section.multiallelic_types: 37 --site-win ${sec_default.site_win}
38 #set $mtype = $section.multiallelic_types
39 #end if
40 --multiallelics "${section.multiallelics}${mtype}"
41 #end if
42 ${section.strict_filter}
43 #if $section.site_win:
44 --site-win "${section.site_win}"
45 #end if 38 #end if
46 39
47 #set $section = $sec_restrict 40 #set $section = $sec_restrict
48 @REGIONS@ 41 @REGIONS@
49 @TARGETS@ 42 @TARGETS@
56 > '$output_file' 49 > '$output_file'
57 ]]> 50 ]]>
58 </command> 51 </command>
59 <inputs> 52 <inputs>
60 <expand macro="macro_input" /> 53 <expand macro="macro_input" />
61 <section name="sec_restrict" expanded="false" title="Restrict to"> 54 <expand macro="macro_fasta_ref" />
62 <expand macro="macro_regions" /> 55 <param name="check_ref" type="select" display="radio"
63 <expand macro="macro_targets" /> 56 label="When any REF allele does not match the reference genome base"
64 </section> 57 help="Warnings about REF mismatches will be emitted to the standard
65 <section name="sec_ref" expanded="false" title="Reference Options"> 58 error (stderr) stream, and it is recommended to check there for
66 <expand macro="macro_fasta_ref" /> 59 problems if you choose not to exit with an error immediately upon
67 <param name="check_ref" type="select" label="Check REF" optional="true" 60 encountering a mismatch.">
68 help="What to do when incorrect or missing REF allele is encountered"> 61 <option value="w">ignore the problem (-w)</option>
69 <option value="e">check REF alleles and exit (e)</option> 62 <option value="wx">exclude the variant record from the output (-wx)</option>
70 <option value="w">warn (w)</option> 63 <option value="ws">fix the variant record using the reference genome information (-ws)</option>
71 <option value="x">exclude (x)</option> 64 <option value="e">exit with an error (-e)</option>
72 <option value="s">set bad sites (s)</option> 65 </param>
73 <option value="wx">warn and exclude (wx)</option> 66 <param argument="--do-not-normalize" name="normalize_indels" type="boolean" truevalue="" falsevalue="--do-not-normalize" checked="false"
74 <option value="ws">warn and set bad sites (ws)</option> 67 label="Left-align and normalize indels?" />
75 </param> 68 <param name="rm_dup" type="select" display="radio"
76 <param name="do_not_normalize" type="boolean" truevalue="--do-not-normalize" falsevalue="" label="Skip Normalization, just set REF allele" 69 label="Perform deduplication for the folowing types of variant records">
77 help="Do not normalize indels, just fix or set the REF allele from the reference (-N, --do-not-normalize with -m or -c s)" /> 70 <option value="">do not deduplicate any records</option>
78 </section> 71 <option value="snps">snps</option>
79 72 <option value="indels">indels</option>
80 <section name="sec_default" expanded="false" title="Default Options"> 73 <option value="both">both</option>
81 <param name="rm_dup" type="select" label="Rm Dup" optional="true" 74 <option value="any">any</option>
82 help="Remove duplicate snps|indels|both|any"> 75 </param>
83 <option value="snps">snps</option> 76 <conditional name="multiallelics">
84 <option value="indels">indels</option> 77 <param name="mode" type="select" label="~multiallelics">
85 <option value="both">both</option> 78 <option value="">preserve multiallelic/biallelic sites</option>
86 <option value="any">any</option>
87 </param>
88 <param name="multiallelics" type="select" label="~multiallelics" optional="true">
89 <option value="-">split multiallelic sites into biallelic records (-)</option> 79 <option value="-">split multiallelic sites into biallelic records (-)</option>
90 <option value="+">join biallelic sites into multiallelic records (+)</option> 80 <option value="+">join biallelic sites into multiallelic records (+)</option>
91 </param> 81 </param>
92 <param name="multiallelic_types" type="select" label="variant types which should be split or merged" optional="true"> 82 <when value="" />
93 <option value="snps">snps - only SNP records should be split or merged</option> 83 <when value="-">
94 <option value="indels">indels - only indel records should be split or merged</option> 84 <param name="multiallelic_types" type="select" display="radio"
95 <option value="both">both - records should be split or merged</option> 85 label="split the following variant types">
96 <option value="any">any - SNPs and indels should be merged into a single record</option> 86 <option value="snps">SNPs</option>
97 </param> 87 <option value="indels">indels</option>
98 <param name="strict_filter" type="boolean" truevalue="--strict-filter" falsevalue="" label="Strict Filter" 88 <option value="both" selected="true">both</option>
99 help="When merging (-m+), merged site is PASS only if all sites being merged PASS" /> 89 </param>
90 </when>
91 <when value="+">
92 <param name="multiallelic_types" type="select" display="radio"
93 label="merge the following variant types">
94 <option value="snps">SNPs</option>
95 <option value="indels">indels</option>
96 <option value="both" selected="true">SNPs and indels, but keep variants of the two types separate (both)</option>
97 <option value="any">SNPs and indels, and merge variant records of different types (any)</option>
98 </param>
99 <param name="strict_filter" type="boolean" truevalue="--strict-filter" falsevalue=""
100 label="Strict Filter"
101 help="merged site is PASS only if all sites being merged PASS" />
102 </when>
103 </conditional>
104 <section name="sec_restrict" expanded="false" title="Restrict all operations to">
105 <expand macro="macro_restrict" type="region" label_type="Region" />
106 <expand macro="macro_restrict" type="target" label_type="Target">
107 <expand macro="macro_invert_targets" />
108 </expand>
109 </section>
110
111 <section name="sec_default" expanded="false" title="Other Options">
100 <param name="site_win" type="integer" label="Site Window" value="1000" optional="True" 112 <param name="site_win" type="integer" label="Site Window" value="1000" optional="True"
101 help="(-w, --site-win) Buffer for sorting lines which changed position during realignment" /> 113 help="(-w, --site-win) Buffer for sorting lines which changed position during realignment" />
102 </section> 114 </section>
103
104 <expand macro="macro_select_output_type" /> 115 <expand macro="macro_select_output_type" />
105 </inputs> 116 </inputs>
106 <outputs> 117 <outputs>
107 <expand macro="macro_vcf_output"/> 118 <expand macro="macro_vcf_output"/>
108 </outputs> 119 </outputs>
109 <tests> 120 <tests>
110 <test> 121 <test>
111 <param name="input_file" ftype="vcf" value="norm.vcf" /> 122 <param name="input_file" ftype="vcf" value="norm.vcf" />
112 <param name="fasta_ref" ftype="fasta" value="norm.fa" /> 123 <expand macro="test_using_reference" ref="norm.fa" />
124 <param name="normalize_indels" value="true" />
113 <param name="output_type" value="v" /> 125 <param name="output_type" value="v" />
114 <output name="output_file"> 126 <output name="output_file">
115 <assert_contents> 127 <assert_contents>
116 <has_text text="T,TAACCCTA" /> 128 <has_text text="T,TAACCCTA" />
117 <not_has_text text="TAA,TAACCCTAAA" /> 129 <not_has_text text="TAA,TAACCCTAAA" />
118 </assert_contents> 130 </assert_contents>
119 </output> 131 </output>
120 </test> 132 </test>
121 133
122 <test> 134 <test>
135 <param name="input_file" ftype="vcf" dbkey="?" value="norm.vcf" />
136 <expand macro="test_using_reference" select_from="cached" ref="norm" />
137 <param name="normalize_indels" value="true" />
138 <param name="output_type" value="v" />
139 <output name="output_file">
140 <assert_contents>
141 <has_text text="T,TAACCCTA" />
142 <not_has_text text="TAA,TAACCCTAAA" />
143 </assert_contents>
144 </output>
145 </test>
146
147 <test>
123 <param name="input_file" ftype="vcf" value="norm.split.vcf" /> 148 <param name="input_file" ftype="vcf" value="norm.split.vcf" />
124 <param name="multiallelics" value="-" /> 149 <expand macro="test_using_reference" ref="norm.fa" />
150 <conditional name="multiallelics">
151 <param name="mode" value="-" />
152 </conditional>
125 <param name="output_type" value="v" /> 153 <param name="output_type" value="v" />
126 <output name="output_file"> 154 <output name="output_file">
127 <assert_contents> 155 <assert_contents>
128 <not_has_text text="TAA,TAACCCTAAA" /> 156 <not_has_text text="TAA,TAACCCTAAA" />
129 <has_text_matching expression="1\t105\t.\tTAAACCCTAAA\tTAA\t"/> 157 <has_text_matching expression="1\t105\t.\tTAAACCCTAAA\tTAA\t"/>
132 </output> 160 </output>
133 </test> 161 </test>
134 162
135 <test> 163 <test>
136 <param name="input_file" ftype="vcf" value="norm.split.vcf" /> 164 <param name="input_file" ftype="vcf" value="norm.split.vcf" />
137 <param name="fasta_ref" ftype="fasta" value="norm.fa" /> 165 <expand macro="test_using_reference" ref="norm.fa" />
138 <param name="multiallelics" value="-" /> 166 <param name="normalize_indels" value="true" />
167 <conditional name="multiallelics">
168 <param name="mode" value="-" />
169 </conditional>
139 <param name="output_type" value="v" /> 170 <param name="output_type" value="v" />
140 <output name="output_file"> 171 <output name="output_file">
141 <assert_contents> 172 <assert_contents>
142 <not_has_text text="TAA,TAACCCTAAA" /> 173 <not_has_text text="TAA,TAACCCTAAA" />
143 <has_text_matching expression="1\t105\t.\tTAAACCCTA\tT\t"/> 174 <has_text_matching expression="1\t105\t.\tTAAACCCTA\tT\t"/>
146 </output> 177 </output>
147 </test> 178 </test>
148 179
149 <test> 180 <test>
150 <param name="input_file" ftype="vcf" value="norm.merge.vcf" /> 181 <param name="input_file" ftype="vcf" value="norm.merge.vcf" />
151 <param name="multiallelics" value="+" /> 182 <expand macro="test_using_reference" ref="norm.fa" />
183 <conditional name="multiallelics">
184 <param name="mode" value="+" />
185 </conditional>
152 <param name="output_type" value="v" /> 186 <param name="output_type" value="v" />
153 <output name="output_file"> 187 <output name="output_file">
154 <assert_contents> 188 <assert_contents>
155 <has_text text="TAA,TAACCCTAAA" /> 189 <has_text text="TAA,TAACCCTAAA" />
156 <has_text_matching expression="2\t114\t.\tTC\tTTCC,TTC\t999\tFAIL"/> 190 <has_text_matching expression="2\t114\t.\tTC\tTTCC,TTC\t999\tFAIL"/>
157 </assert_contents> 191 </assert_contents>
158 </output> 192 </output>
159 </test> 193 </test>
160 <test> 194 <test>
161 <param name="input_file" ftype="vcf" value="norm.merge.vcf" /> 195 <param name="input_file" ftype="vcf" value="norm.merge.vcf" />
162 <param name="multiallelics" value="+" /> 196 <expand macro="test_using_reference" ref="norm.fa" />
163 <param name="strict_filter" value="true" /> 197 <conditional name="multiallelics">
198 <param name="mode" value="+" />
199 <param name="strict_filter" value="true" />
200 </conditional>
164 <param name="output_type" value="v" /> 201 <param name="output_type" value="v" />
165 <output name="output_file"> 202 <output name="output_file">
166 <assert_contents> 203 <assert_contents>
167 <has_text text="TAA,TAACCCTAAA" /> 204 <has_text text="TAA,TAACCCTAAA" />
168 <has_text_matching expression="2\t114\t.\tTC\tTTCC,TTC\t999\tPASS"/> 205 <has_text_matching expression="2\t114\t.\tTC\tTTCC,TTC\t999\tPASS"/>
170 </output> 207 </output>
171 </test> 208 </test>
172 209
173 <test> 210 <test>
174 <param name="input_file" ftype="vcf" value="norm.setref.vcf" /> 211 <param name="input_file" ftype="vcf" value="norm.setref.vcf" />
175 <param name="fasta_ref" ftype="fasta" value="norm.fa" /> 212 <expand macro="test_using_reference" ref="norm.fa" />
176 <param name="do_not_normalize" value="true" /> 213 <param name="check_ref" value="ws" />
177 <param name="check_ref" value="s" />
178 <param name="output_type" value="v" /> 214 <param name="output_type" value="v" />
179 <output name="output_file"> 215 <output name="output_file">
180 <assert_contents> 216 <assert_contents>
181 <has_text_matching expression="2\t101\t.\tA\tc\t999\tPASS"/> 217 <has_text_matching expression="2\t101\t.\tA\tc\t999\tPASS"/>
182 <has_text_matching expression="2\t105\t.\tT\t&lt;DEL&gt;\t999\tPASS"/> 218 <has_text_matching expression="2\t105\t.\tT\t&lt;DEL&gt;\t999\tPASS"/>