Mercurial > repos > iuc > medaka_variant

comparison variant.xml @ 13:b7da0d412e48 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/medaka commit 86211daa63a6f39524df8759364795b782324303"
author iuc
date Fri, 17 Sep 2021 20:20:30 +0000
parents 8ee6b81b7de9
children 8ba58c8139aa
comparison
equal deleted inserted replaced
12:3d39a182b8e7 13:b7da0d412e48
1 <?xml version="1.0"?> 1 <?xml version="1.0"?>
2 <tool id="medaka_variant" name="medaka variant tool" version="@TOOL_VERSION@+galaxy0" profile="@PROFILE@"> 2 <tool id="medaka_variant" name="medaka variant tool" version="@TOOL_VERSION@+galaxy1" profile="@PROFILE@">
3 <description>Probability decoding</description> 3 <description>decodes variant calls from medaka consensus output</description>
4 <macros> 4 <macros>
5 <import>macros.xml</import> 5 <import>macros.xml</import>
6 </macros> 6 </macros>
7 <expand macro="requirements"/> 7 <expand macro="requirements"/>
8
9 <expand macro="version_command"/> 8 <expand macro="version_command"/>
10
11 <configfiles>
12 <configfile name="convert_fasta">
13 import sys
14 infile = open(sys.argv[1], 'r')
15 outfile = open(sys.argv[2], 'w')
16 for line in infile:
17 if line[0] == '>':
18 outfile.write(line)
19 else:
20 outfile.write(line.upper())
21 infile.close()
22 outfile.close()
23 </configfile>
24 </configfiles>
25
26 <command detect_errors="exit_code"><![CDATA[ 9 <command detect_errors="exit_code"><![CDATA[
27 ## initialize 10 ## initialize
28 @REF_FASTA@ 11 @REF_FASTA@
29 12
30 13
42 ## required 25 ## required
43 reference.fa 26 reference.fa
44 #for $current in $pool.inputs 27 #for $current in $pool.inputs
45 '$current' 28 '$current'
46 #end for 29 #end for
47 '$out_result' ## output
48 2>&1 | tee '$out_log'
49 #elif $pool.pool_mode == "No": 30 #elif $pool.pool_mode == "No":
50 ## run 31 ## run
51 medaka variant 32 medaka variant
52 ## optional 33 ## optional
53 --debug 34 --debug
58 ${ambig_ref} 39 ${ambig_ref}
59 ${gvcf} 40 ${gvcf}
60 ## required 41 ## required
61 reference.fa 42 reference.fa
62 '$pool.input' 43 '$pool.input'
63 '$out_result' ##output 44 #end if
45 #if str($output_annotated.output_annotated_select) == 'false':
46 '$out_variants' ##output
64 2>&1 | tee '$out_log' 47 2>&1 | tee '$out_log'
65 #end if 48 #else
66 #if $out_annotated: 49 raw.vcf ##output of medaka variant
67 ## medaka annotate errors out if the reference is lower case at a position it's annotating because it checks vs the ref base in the vcf 50 2>&1 | tee '$out_log'
68 && python '$convert_fasta' reference.fa upper_reference.fa
69 && ln -s '$output_annotated.in_bam' in.bam 51 && ln -s '$output_annotated.in_bam' in.bam
70 && ln -s '$output_annotated.in_bam.metadata.bam_index' in.bai 52 && ln -s '$output_annotated.in_bam.metadata.bam_index' in.bai
71 && medaka tools annotate --dpsp --pad $output_annotated.pad '$out_result' upper_reference.fa in.bam tmp.vcf 53 && medaka tools annotate --dpsp --pad $output_annotated.pad raw.vcf reference.fa in.bam tmp.vcf
72 && python '$__tool_directory__/convert_VCF_info_fields.py' tmp.vcf '$out_annotated' 54 && python '$__tool_directory__/convert_VCF_info_fields.py' tmp.vcf '$out_variants'
73 #end if 55 #end if
74 ]]></command> 56 ]]></command>
75 <inputs> 57 <inputs>
76 <conditional name="pool"> 58 <conditional name="pool">
77 <param name="pool_mode" type="select" label="Are you pooling HDF5 datasets?"> 59 <param name="pool_mode" type="select" label="Are you pooling HDF5 datasets?">
97 </param> 79 </param>
98 <param argument="--verbose" type="boolean" truevalue="--verbose" falsevalue="" label="Populate VCF info fields?"/> 80 <param argument="--verbose" type="boolean" truevalue="--verbose" falsevalue="" label="Populate VCF info fields?"/>
99 <param argument="--ambig_ref" type="boolean" truevalue="--ambig_ref" falsevalue="" label="Decode variants at ambiguous reference positions?" checked="false"/> 81 <param argument="--ambig_ref" type="boolean" truevalue="--ambig_ref" falsevalue="" label="Decode variants at ambiguous reference positions?" checked="false"/>
100 <param argument="--gvcf" type="boolean" truevalue="--gvcf" falsevalue="" label="Output VCF records for reference loci predicted to be non-variant?" checked="false"/> 82 <param argument="--gvcf" type="boolean" truevalue="--gvcf" falsevalue="" label="Output VCF records for reference loci predicted to be non-variant?" checked="false"/>
101 <conditional name="output_annotated"> 83 <conditional name="output_annotated">
102 <param name="output_annotated_select" type="select" label="Output annotated VCF?" help="Annotate allele frequency, depth of coverage, etc for each variant (requires BAM file)"> 84 <param name="output_annotated_select" type="select"
103 <option value="true" selected="true">Output annotated VCF</option> 85 label="Type of VCF to generate"
104 <option value="false">Don't output annotated VCF</option> 86 help="Variant INFO fields in the VCF can be extended to include allele frequency, depth of coverage, etc., but this requires a BAM dataset to calculate those values from.">
87 <option value="true" selected="true">Write annotated VCF with extended INFO</option>
88 <option value="false">Write original decoded VCF with minimal INFO field</option>
105 </param> 89 </param>
106 <when value="true"> 90 <when value="true">
107 <param name="in_bam" type="data" format="bam" optional="false" label="BAM to annotate the VCF"/> 91 <param name="in_bam" type="data" format="bam" optional="false" label="BAM to caclulate additional INFO fields from"/>
108 <param name="pad" type="integer" min="1" value="25" label="Padding width on either side of variant for realignment in medaka tools anntotate, used to calculate DPSP, DPSPS, AF, FAF, RAF, SB, DP4, and AS in the output annotated VCF"/> 92 <param name="pad" type="integer" min="1" value="25"
93 label="Padding width on either side of variant for realignment"
94 help="To calculate the additional INFO fields the tool will run medaka tools anntotate, which performs local realignment of the region +- this width around each variant. All calculated new fields will depend on the width chosen, so only change this value if you know what you are doing." />
109 </when> 95 </when>
110 <when value="false"/> 96 <when value="false"/>
111 </conditional> 97 </conditional>
112 <param name="output_log_bool" type="boolean" label="Output log file?" checked="true"/> 98 <param name="output_log_bool" type="boolean" label="Output log file?" checked="true"/>
113 </inputs> 99 </inputs>
114 <outputs> 100 <outputs>
115 <!-- standard --> 101 <data name="out_variants" format="vcf" label="${tool.name} on ${on_string}: called variants"/>
116 <data name="out_result" format="vcf" label="${tool.name} on ${on_string}: Result"/>
117 <!-- optional -->
118 <data name="out_annotated" format="vcf" label="${tool.name} on ${on_string}: Annotated">
119 <filter>output_annotated['output_annotated_select']!='false'</filter>
120 </data>
121 <data name="out_log" format="tabular" label="${tool.name} on ${on_string}: Log"> 102 <data name="out_log" format="tabular" label="${tool.name} on ${on_string}: Log">
122 <filter>output_log_bool</filter> 103 <filter>output_log_bool</filter>
123 </data> 104 </data>
124 </outputs> 105 </outputs>
125 <tests> 106 <tests>
126 <!-- #1 default --> 107 <!-- #1 default -->
127 <test expect_num_outputs="3"> 108 <test expect_num_outputs="2">
128 <conditional name="pool"> 109 <conditional name="pool">
129 <param name="pool_mode" value="Yes"/> 110 <param name="pool_mode" value="Yes"/>
130 <param name="inputs" value="medaka_test.hdf,medaka_test.hdf"/> 111 <param name="inputs" value="medaka_test.hdf,medaka_test.hdf"/>
131 </conditional> 112 </conditional>
132 <conditional name="reference_source"> 113 <conditional name="reference_source">
133 <param name="reference_source_selector" value="history"/> 114 <param name="reference_source_selector" value="history"/>
134 <param name="ref_file" value="ref.fasta.gz"/> 115 <param name="ref_file" value="ref.fasta"/>
135 </conditional> 116 </conditional>
136 <param name="ambig_ref" value="true"/> 117 <param name="ambig_ref" value="true"/>
137 <conditional name="output_annotated"> 118 <conditional name="output_annotated">
138 <param name="output_annotated_select" value="true"/> 119 <param name="output_annotated_select" value="true"/>
139 <param name="in_bam" value="medaka_test.bam"/> 120 <param name="in_bam" value="medaka_test.bam"/>
140 </conditional> 121 </conditional>
141 <param name="output_log_bool" value="true"/> 122 <param name="output_log_bool" value="true"/>
142 123 <output name="out_variants">
143 <output name="out_result"> 124 <assert_contents>
144 <assert_contents> 125 <has_n_lines n="18"/>
145 <has_n_lines n="9"/>
146 <has_line line="##fileformat=VCFv4.1" />
147 <has_line_matching expression="##medaka_version=[0-9]+\.[0-9]+\.[0-9]+" />
148 </assert_contents>
149 </output>
150 <output name="out_annotated">
151 <assert_contents>
152 <has_n_lines n="23"/>
153 <has_line line="##fileformat=VCFv4.1" /> 126 <has_line line="##fileformat=VCFv4.1" />
154 <has_line_matching expression="##medaka_version=[0-9]+\.[0-9]+\.[0-9]+" /> 127 <has_line_matching expression="##medaka_version=[0-9]+\.[0-9]+\.[0-9]+" />
155 <has_line_matching expression="##convert_VCF_info_fields=[0-9]+\.[0-9]+" /> 128 <has_line_matching expression="##convert_VCF_info_fields=[0-9]+\.[0-9]+" />
156 </assert_contents> 129 </assert_contents>
157 </output> 130 </output>
160 <has_n_lines n="10" /> 133 <has_n_lines n="10" />
161 </assert_contents> 134 </assert_contents>
162 </output> 135 </output>
163 </test> 136 </test>
164 <!--No pooling--> 137 <!--No pooling-->
165 <test expect_num_outputs="3"> 138 <test expect_num_outputs="2">
166 <conditional name="pool"> 139 <conditional name="pool">
167 <param name="pool_mode" value="No"/> 140 <param name="pool_mode" value="No"/>
168 <param name="input" value="medaka_test.hdf"/> 141 <param name="input" value="medaka_test.hdf"/>
169 </conditional> 142 </conditional>
170 <conditional name="reference_source"> 143 <conditional name="reference_source">
175 <conditional name="output_annotated"> 148 <conditional name="output_annotated">
176 <param name="output_annotated_select" value="true"/> 149 <param name="output_annotated_select" value="true"/>
177 <param name="in_bam" value="medaka_test.bam"/> 150 <param name="in_bam" value="medaka_test.bam"/>
178 </conditional> 151 </conditional>
179 <param name="output_log_bool" value="true"/> 152 <param name="output_log_bool" value="true"/>
180 153 <output name="out_variants">
181 <output name="out_result"> 154 <assert_contents>
182 <assert_contents> 155 <has_n_lines n="18"/>
183 <has_n_lines n="9"/>
184 <has_line line="##fileformat=VCFv4.1" />
185 <has_line_matching expression="##medaka_version=[0-9]+\.[0-9]+\.[0-9]+" />
186 </assert_contents>
187 </output>
188 <output name="out_annotated">
189 <assert_contents>
190 <has_n_lines n="23"/>
191 <has_line line="##fileformat=VCFv4.1" /> 156 <has_line line="##fileformat=VCFv4.1" />
192 <has_line_matching expression="##medaka_version=[0-9]+\.[0-9]+\.[0-9]+" /> 157 <has_line_matching expression="##medaka_version=[0-9]+\.[0-9]+\.[0-9]+" />
193 <has_line_matching expression="##convert_VCF_info_fields=[0-9]+\.[0-9]+" /> 158 <has_line_matching expression="##convert_VCF_info_fields=[0-9]+\.[0-9]+" />
194 </assert_contents> 159 </assert_contents>
195 </output> 160 </output>
211 </conditional> 176 </conditional>
212 <conditional name="output_annotated"> 177 <conditional name="output_annotated">
213 <param name="output_annotated_select" value="false"/> 178 <param name="output_annotated_select" value="false"/>
214 </conditional> 179 </conditional>
215 <param name="output_log_bool" value="false"/> 180 <param name="output_log_bool" value="false"/>
216 181 <output name="out_variants">
217 <output name="out_result">
218 <assert_contents> 182 <assert_contents>
219 <has_n_lines n="9"/> 183 <has_n_lines n="9"/>
220 <has_line line="##fileformat=VCFv4.1" /> 184 <has_line line="##fileformat=VCFv4.1" />
221 <has_line_matching expression="##medaka_version=[0-9]+\.[0-9]+\.[0-9]+" /> 185 <has_line_matching expression="##medaka_version=[0-9]+\.[0-9]+\.[0-9]+" />
222 </assert_contents> 186 </assert_contents>