Mercurial > repos > greg > vsnp_get_snps

comparison vsnp_get_snps.xml @ 8:5e4595b9f63c draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/gregvonkuster/galaxy_tools/tree/master/tools/sequence_analysis/vsnp/vsnp_get_snps commit 7423e5bb852a786195c095b9f663aac0ec9c8fd9"
author greg
date Thu, 29 Jul 2021 12:50:01 +0000
parents 2286f3a13e4d
children 0fe292b20b9d
comparison
equal deleted inserted replaced
7:2286f3a13e4d 8:5e4595b9f63c
1 <tool id="vsnp_get_snps" name="vSNP: get SNPs" version="@WRAPPER_VERSION@.0+galaxy0" profile="@PROFILE@"> 1 <tool id="vsnp_get_snps_broken" name="vSNP: get SNPs broken" version="@WRAPPER_VERSION@.0+galaxy0" profile="@PROFILE@">
2 <description></description> 2 <description></description>
3 <macros> 3 <macros>
4 <import>macros.xml</import> 4 <import>macros.xml</import>
5 </macros> 5 </macros>
6 <requirements> 6 <requirements>
10 <requirement type="package" version="2.0.1">xlrd</requirement> 10 <requirement type="package" version="2.0.1">xlrd</requirement>
11 </requirements> 11 </requirements>
12 <command detect_errors="exit_code"><![CDATA[ 12 <command detect_errors="exit_code"><![CDATA[
13 #import re 13 #import re
14 14
15 #if str($input.metadata.dbkey) == '?':
16 >&2 echo "The dbkey must be set for the zero coverage VCF files with SNPs found in closely related isolate groups" && exit 1
17 #end if
18
15 #set input_vcf_dir = 'input_vcf_dir' 19 #set input_vcf_dir = 'input_vcf_dir'
16 #set output_json_avg_mq_dir = 'output_json_avg_mq_dir' 20 #set output_json_avg_mq_dir = 'output_json_avg_mq_dir'
17 #set output_json_snps_dir = 'output_json_snps_dir' 21 #set output_json_snps_dir = 'output_json_snps_dir'
18 #set output_snps_dir = 'output_snps_dir' 22 #set output_snps_dir = 'output_snps_dir'
19 23
20 mkdir -p $input_vcf_dir && 24 mkdir -p $input_vcf_dir &&
21 mkdir -p $output_json_avg_mq_dir && 25 mkdir -p $output_json_avg_mq_dir &&
22 mkdir -p $output_json_snps_dir && 26 mkdir -p $output_json_snps_dir &&
23 mkdir -p $output_snps_dir && 27 mkdir -p $output_snps_dir &&
24 28
25 #set dbkey = '?' 29 #set input_identifier = re.sub('[^\s\w\-]', '_', str($input.element_identifier))
26 #for $i in $input_vcf_collection: 30 ln -s '${input}' '$input_vcf_dir/${input_identifier}' &&
27 #if str($dbkey) == '?': 31
28 #set dbkey = $i.metadata.dbkey
29 #else if str($dbkey) != $i.metadata.dbkey:
30 >&2 echo "The dbkeys associated with the zero coverage VCF files with SNPs found in closely related isolate groups are not unique" &&
31 exit 1
32 #end if
33 #set vcf_identifier = re.sub('[^\s\w\-]', '_', str($i.element_identifier))
34 ln -s '${i}' '$input_vcf_dir/${vcf_identifier}' &&
35 #end for
36 #if str($dbkey) == '?':
37 >&2 echo "The dbkey must be set for the zero coverage VCF files with SNPs found in closely related isolate groups" && exit 1
38 #end if
39 #if str($input_zc_vcf_type_cond.input_zc_vcf_type) == "single": 32 #if str($input_zc_vcf_type_cond.input_zc_vcf_type) == "single":
40 #set zc_vcf_identifier = re.sub('[^\s\w\-]', '_', str($input_zc_vcf.element_identifier)) 33 #set zc_vcf_identifier = re.sub('[^\s\w\-]', '_', str($input_zc_vcf.element_identifier))
41 ln -s '${input_zc_vcf}' '$input_vcf_dir/${zc_vcf_identifier}' && 34 ln -s '${input_zc_vcf}' '$input_vcf_dir/${zc_vcf_identifier}' &&
42 #else 35 #else
43 #for $i in $input_zc_vcf_type_cond.input_zc_vcf_collection: 36 #for $i in $input_zc_vcf_type_cond.input_zc_vcf_collection:
50 #set excel_file = 'No genome specified for input VCF (database) file(s)' 43 #set excel_file = 'No genome specified for input VCF (database) file(s)'
51 #set excel_fields = $__app__.tool_data_tables['vsnp_excel'].get_fields() 44 #set excel_fields = $__app__.tool_data_tables['vsnp_excel'].get_fields()
52 ## The value of excel_fields is a nested list that looks like this. 45 ## The value of excel_fields is a nested list that looks like this.
53 ## [['AF2122', 'Mbovis_define_filter.xlsx', '~/tool-data/vsnp/AF2122/excel/Mbovis_define_filter.xlsx', 'Excel file for AF2122'],...] 46 ## [['AF2122', 'Mbovis_define_filter.xlsx', '~/tool-data/vsnp/AF2122/excel/Mbovis_define_filter.xlsx', 'Excel file for AF2122'],...]
54 #for $i in $excel_fields: 47 #for $i in $excel_fields:
55 #if str($i[0]) == $dbkey: 48 #if str($i[0]) == $input.metadata.dbkey:
56 #set excel_file = $i[2] 49 #set excel_file = $i[2]
57 #break 50 #break
58 #end if 51 #end if
59 #end for 52 #end for
60 #else: 53 #else:
74 --output_json_snps_dir '$output_json_snps_dir' 67 --output_json_snps_dir '$output_json_snps_dir'
75 --output_snps_dir '$output_snps_dir' 68 --output_snps_dir '$output_snps_dir'
76 --output_summary '$output_summary' 69 --output_summary '$output_summary'
77 --processes \${GALAXY_SLOTS:-8} 70 --processes \${GALAXY_SLOTS:-8}
78 --quality_score_n_threshold $quality_score_n_threshold 71 --quality_score_n_threshold $quality_score_n_threshold
79 --dbkey '$dbkey' 72 --dbkey '$input.metadata.dbkey'
80 ]]></command> 73 ]]></command>
81 <inputs> 74 <inputs>
82 <conditional name="input_zc_vcf_type_cond"> 75 <conditional name="input_zc_vcf_type_cond">
83 <param name="input_zc_vcf_type" type="select" label="Choose the category of the files to be analyzed"> 76 <param name="input_zc_vcf_type" type="select" label="Choose the category of the files to be analyzed">
84 <option value="collection" selected="true">A collection of zero coverage VCF files</option> 77 <option value="collection" selected="true">A collection of zero coverage VCF files</option>
89 </when> 82 </when>
90 <when value="collection"> 83 <when value="collection">
91 <param name="input_zc_vcf_collection" format="vcf" type="data_collection" collection_type="list" label="Collection of zero coverage VCF files"/> 84 <param name="input_zc_vcf_collection" format="vcf" type="data_collection" collection_type="list" label="Collection of zero coverage VCF files"/>
92 </when> 85 </when>
93 </conditional> 86 </conditional>
94 <param name="input_vcf_collection" format="vcf" type="data_collection" collection_type="list" label="Collection of zero coverage VCF files with SNPs found in closely related isolate groups"/> 87 <param name="input" format="vcf" type="data" label="Zero coverage VCF file with SNPs found in closely related isolate groups"/>
95 <param name="ac" type="integer" min="0" value="2" label="Allele count threshold" help="At least 1 position must have this value for a SNP to be added to a group"/> 88 <param name="ac" type="integer" min="0" value="2" label="Allele count threshold" help="At least 1 position must have this value for a SNP to be added to a group"/>
96 <param name="min_mq" type="integer" min="0" value="56" label="Map quality threshold" help="At least 1 position must have a higher MQ value for a SNP to be added to a group"/> 89 <param name="min_mq" type="integer" min="0" value="56" label="Map quality threshold" help="At least 1 position must have a higher MQ value for a SNP to be added to a group"/>
97 <param name="min_quality_score" type="integer" min="0" value="150" label="Quality score threshold" help="At least 1 position must have a higher quality score for a SNP to be added to a group"/> 90 <param name="min_quality_score" type="integer" min="0" value="150" label="Quality score threshold" help="At least 1 position must have a higher quality score for a SNP to be added to a group"/>
98 <param name="quality_score_n_threshold" type="integer" min="0" value="150" label="Minimum quality score N value for alleles" help="Alleles are marked as N for quality scores between this value and the minimum quality score value above"/> 91 <param name="quality_score_n_threshold" type="integer" min="0" value="150" label="Minimum quality score N value for alleles" help="Alleles are marked as N for quality scores between this value and the minimum quality score value above"/>
99 <conditional name="input_excel_cond"> 92 <conditional name="input_excel_cond">
108 <option value="history">from history</option> 101 <option value="history">from history</option>
109 </param> 102 </param>
110 <when value="cached"> 103 <when value="cached">
111 <param name="input_excel" type="select" label="Excel file"> 104 <param name="input_excel" type="select" label="Excel file">
112 <options from_data_table="vsnp_excel"> 105 <options from_data_table="vsnp_excel">
113 <filter type="data_meta" column="0" key="dbkey" ref="input_vcf_collection"/>
114 <validator type="no_options" message="No built-in Excel grouping and filtering datasets are available"/> 106 <validator type="no_options" message="No built-in Excel grouping and filtering datasets are available"/>
115 </options> 107 </options>
116 </param> 108 </param>
117 </when> 109 </when>
118 <when value="history"> 110 <when value="history">
135 <discover_datasets pattern="__name_and_ext__" directory="output_json_snps_dir"/> 127 <discover_datasets pattern="__name_and_ext__" directory="output_json_snps_dir"/>
136 </collection> 128 </collection>
137 <data name="output_summary" format="html" label="${tool.name} on ${on_string} (summary)"/> 129 <data name="output_summary" format="html" label="${tool.name} on ${on_string} (summary)"/>
138 </outputs> 130 </outputs>
139 <tests> 131 <tests>
140 <!-- 132 <!-- No excel file, all_isolates is False -->
141 Unfortunately the test files cannot be gzipped since Galaxy changes the file names
142 to be something like 00-0121_WI_Cervid_99-A_vcf_gz, and the VCF Reader requires
143 gzipped files to have a .gz extension. The exception is
144 UnicodeDecodeError: 'utf-8' codec can't decode byte 0x8b in position 1: invalid start byte
145 -->
146 <!-- A single vcf input, no excel file, all_isolates is False -->
147 <test expect_num_outputs="4"> 133 <test expect_num_outputs="4">
148 <param name="input_zc_vcf_type" value="single"/> 134 <param name="input" value="input_zc_vcf.vcf" ftype="vcf" dbkey="89"/>
149 <param name="input_zc_vcf" value="input_zc_vcf.vcf" ftype="vcf" dbkey="89"/> 135 <param name="input_zc_vcf_collection">
150 <param name="input_vcf_collection">
151 <collection type="list"> 136 <collection type="list">
152 <element name="SRR8073662_zc.vcf" value="SRR8073662_zc.vcf" dbkey="89"/> 137 <element name="SRR8073662_zc.vcf" value="SRR8073662_zc.vcf" dbkey="89"/>
153 <element name="SRR1792272_zc.vcf" value="SRR1792272_zc.vcf" dbkey="89"/> 138 <element name="SRR1792272_zc.vcf" value="SRR1792272_zc.vcf" dbkey="89"/>
154 </collection> 139 </collection>
155 </param> 140 </param>
156 <param name="input_excel_param" value="no"/> 141 <param name="input_excel_param" value="no"/>
157 <output_collection name="snps" type="list" count="1"> 142 <output_collection name="snps" type="list" count="1">
158 <element name="all_vcf" file="all_vcf.fasta" ftype="fasta" compare="contains"/> 143 <element name="all_vcf" file="all_vcf.fasta" ftype="fasta" compare="contains"/>
159 </output_collection> 144 </output_collection>
160 <output_collection name="json_avg_mq" type="list" count="1"> 145 <output_collection name="json_avg_mq" type="list" count="1">
161 <element name="all_vcf" file="json_avg_mq_all_vcf.json" ftype="json" compare="contains"/> 146 <element name="all_vcf" file="avg_mq.json" ftype="json" compare="contains"/>
162 </output_collection> 147 </output_collection>
163 <output_collection name="json_snps" type="list" count="1"> 148 <output_collection name="json_snps" type="list" count="1">
164 <element name="all_vcf" file="json_all_vcf.json" ftype="json" compare="contains"/> 149 <element name="all_vcf" file="json_all_vcf.json" ftype="json" compare="contains"/>
165 </output_collection> 150 </output_collection>
166 <output name="output_summary" file="output_summary.html" ftype="html" compare="contains"/> 151 <output name="output_summary" file="output_summary.html" ftype="html" compare="contains"/>
167 </test> 152 </test>
168 <!-- An input collection, no excel file, all_isolates is False --> 153 <!-- Excel file, all_isolates is False -->
169 <test expect_num_outputs="4"> 154 <test expect_num_outputs="4">
170 <param name="input_zc_vcf_type" value="collection"/> 155 <param name="input" value="02_0585_COA_TX_Fed.vcf" ftype="vcf" dbkey="89"/>
171 <param name="input_zc_vcf_collection"> 156 <param name="input_zc_vcf_collection">
172 <collection type="list"> 157 <collection type="list">
173 <element name="BCG_Pasteur_Unknown_FR_SRR8886989.vcf" value="BCG_Pasteur_Unknown_FR_SRR8886989.vcf" dbkey="89"/> 158 <element name="L02-19-1908_zc.vcf" value="L02-19-1908_zc.vcf" dbkey="89"/>
174 <element name="BCG_Tokyo_Unknown_JP_DRR029468.vcf" value="BCG_Tokyo_Unknown_JP_DRR029468.vcf" dbkey="89"/> 159 <element name="L03-13-2334_zc.vcf" value="L03-13-2334_zc.vcf" dbkey="89"/>
175 </collection> 160 <element name="L03-16-0255_zc.vcf" value="L03-16-0255_zc.vcf" dbkey="89"/>
176 </param>
177 <param name="input_vcf_collection">
178 <collection type="list">
179 <element name="01_1787_FL_Zoo_Jaguar.vcf" value="01_1787_FL_Zoo_Jaguar.vcf" dbkey="89"/>
180 <element name="02_5877_MEX_TX_Fed.vcf" value="02_5877_MEX_TX_Fed.vcf" dbkey="89"/>
181 <element name="02_0585_COA_TX_Fed.vcf" value="02_0585_COA_TX_Fed.vcf" dbkey="89"/>
182 </collection>
183 </param>
184 <param name="input_excel_param" value="no"/>
185 <output_collection name="snps" type="list" count="1">
186 <element name="all_vcf" file="all_vcf2.fasta" ftype="fasta" compare="contains"/>
187 </output_collection>
188 <output_collection name="json_avg_mq" type="list" count="1">
189 <element name="all_vcf" file="json_avg_mq_all_vcf.json" ftype="json" compare="contains"/>
190 </output_collection>
191 <output_collection name="json_snps" type="list" count="1">
192 <element name="all_vcf" file="json_all_vcf.json" ftype="json" compare="contains"/>
193 </output_collection>
194 <output name="output_summary" file="output_summary2.html" ftype="html" compare="contains"/>
195 </test>
196 <!-- An input collection, an excel file, all_isolates is False -->
197 <test expect_num_outputs="4">
198 <param name="input_zc_vcf_type" value="collection"/>
199 <param name="input_zc_vcf_collection">
200 <collection type="list">
201 <element name="BCG_Pasteur_Unknown_FR_SRR8886989.vcf" value="BCG_Pasteur_Unknown_FR_SRR8886989.vcf" dbkey="89"/>
202 <element name="BCG_Tokyo_Unknown_JP_DRR029468.vcf" value="BCG_Tokyo_Unknown_JP_DRR029468.vcf" dbkey="89"/>
203 </collection>
204 </param>
205 <param name="input_vcf_collection">
206 <collection type="list">
207 <element name="01_1787_FL_Zoo_Jaguar.vcf" value="01_1787_FL_Zoo_Jaguar.vcf" dbkey="89"/>
208 <element name="02_5877_MEX_TX_Fed.vcf" value="02_5877_MEX_TX_Fed.vcf" dbkey="89"/>
209 <element name="02_0585_COA_TX_Fed.vcf" value="02_0585_COA_TX_Fed.vcf" dbkey="89"/>
210 </collection> 161 </collection>
211 </param> 162 </param>
212 <param name="input_excel_param" value="yes"/> 163 <param name="input_excel_param" value="yes"/>
213 <param name="input_excel" value="89"/> 164 <param name="input_excel" value="89"/>
214 <output_collection name="snps" type="list" count="1"> 165 <output_collection name="snps" type="list" count="1">
215 <element name="Mbovis-17" file="Mbovis-17_snps.fasta" ftype="fasta"/> 166 <element name="Mbovis-TB" file="Mbovis-TB.fasta" ftype="fasta" compare="contains"/>
216 </output_collection> 167 </output_collection>
217 <output_collection name="json_avg_mq" type="list" count="1"> 168 <output_collection name="json_avg_mq" type="list" count="1">
218 <element name="Mbovis-17" file="Mbovis-17_avg_mq_json.json" ftype="json" compare="contains"/> 169 <element name="Mbovis-TB" file="avg_mq.json" ftype="json" compare="contains"/>
219 </output_collection> 170 </output_collection>
220 <output_collection name="json_snps" type="list" count="1"> 171 <output_collection name="json_snps" type="list" count="1">
221 <element name="Mbovis-17" file="Mbovis-17_snps_json.json" ftype="json" compare="contains"/> 172 <element name="Mbovis-TB" file="Mbovis-TB_json.json" ftype="json" compare="contains"/>
222 </output_collection> 173 </output_collection>
223 <output name="output_summary" file="output_summary3.html" ftype="html" compare="contains"/> 174 <output name="output_summary" file="output_summary.html" ftype="html" compare="contains"/>
224 </test> 175 </test>
225 <!-- An input collection, an excel file, all_isolates is True --> 176 <!-- Excel file, all_isolates is True -->
226 <test expect_num_outputs="4"> 177 <test expect_num_outputs="4">
227 <param name="input_zc_vcf_type" value="collection"/> 178 <param name="input" value="02_0585_COA_TX_Fed.vcf" ftype="vcf" dbkey="89"/>
228 <param name="input_zc_vcf_collection"> 179 <param name="input_zc_vcf_collection">
229 <collection type="list"> 180 <collection type="list">
230 <element name="BCG_Pasteur_Unknown_FR_SRR8886989.vcf" value="BCG_Pasteur_Unknown_FR_SRR8886989.vcf" dbkey="89"/> 181 <element name="L02-19-1908_zc.vcf" value="L02-19-1908_zc.vcf" dbkey="89"/>
231 <element name="BCG_Tokyo_Unknown_JP_DRR029468.vcf" value="BCG_Tokyo_Unknown_JP_DRR029468.vcf" dbkey="89"/> 182 <element name="L03-13-2334_zc.vcf" value="L03-13-2334_zc.vcf" dbkey="89"/>
232 </collection> 183 <element name="L03-16-0255_zc.vcf" value="L03-16-0255_zc.vcf" dbkey="89"/>
233 </param>
234 <param name="input_vcf_collection">
235 <collection type="list">
236 <element name="01_1787_FL_Zoo_Jaguar.vcf" value="01_1787_FL_Zoo_Jaguar.vcf" dbkey="89"/>
237 <element name="02_5877_MEX_TX_Fed.vcf" value="02_5877_MEX_TX_Fed.vcf" dbkey="89"/>
238 <element name="02_0585_COA_TX_Fed.vcf" value="02_0585_COA_TX_Fed.vcf" dbkey="89"/>
239 </collection> 184 </collection>
240 </param> 185 </param>
241 <param name="input_excel_param" value="yes"/> 186 <param name="input_excel_param" value="yes"/>
242 <param name="input_excel" value="89"/> 187 <param name="input_excel" value="89"/>
243 <param name="all_isolates" value="--all_isolates"/> 188 <param name="all_isolates" value="--all_isolates"/>
244 <output_collection name="snps" type="list" count="2"> 189 <output_collection name="snps" type="list" count="2">
245 <element name="Mbovis-17" file="Mbovis-17_snps.fasta" ftype="fasta"/> 190 <element name="Mbovis-TB" file="Mbovis-TB.fasta" ftype="fasta" compare="contains"/>
246 <element name="all_vcf" file="all_vcf2.fasta" ftype="fasta" compare="contains"/> 191 <element name="all_vcf" file="all_vcf3.fasta" ftype="fasta" compare="contains"/>
247 </output_collection> 192 </output_collection>
248 <output_collection name="json_avg_mq" type="list" count="2"> 193 <output_collection name="json_avg_mq" type="list" count="2">
249 <element name="Mbovis-17" file="Mbovis-17_avg_mq_json.json" ftype="json" compare="contains"/> 194 <element name="Mbovis-TB" file="avg_mq.json" ftype="json" compare="contains"/>
250 <element name="all_vcf" file="Mbovis-17_avg_mq_json.json" ftype="json" compare="contains"/> 195 <element name="all_vcf" file="avg_mq.json" ftype="json" compare="contains"/>
251 </output_collection> 196 </output_collection>
252 <output_collection name="json_snps" type="list" count="2"> 197 <output_collection name="json_snps" type="list" count="2">
253 <element name="Mbovis-17" file="Mbovis-17_snps_json.json" ftype="json" compare="contains"/> 198 <element name="Mbovis-TB" file="Mbovis-TB_json.json" ftype="json" compare="contains"/>
254 <element name="all_vcf" file="Mbovis-17_snps_json.json" ftype="json" compare="contains"/> 199 <element name="all_vcf" file="all_vcf_snps_json.json" ftype="json" compare="contains"/>
255 </output_collection> 200 </output_collection>
256 <output name="output_summary" file="output_summary4.html" ftype="html" compare="contains"/> 201 <output name="output_summary" file="output_summary.html" ftype="html" compare="contains"/>
257 </test> 202 </test>
258 </tests> 203 </tests>
259 <help> 204 <help>
260 **What it does** 205 **What it does**
261 206