|
1
|
1 <tool id="sniplay_vcftoolsstats" name="VCF tools Stats" version="1.0.0">
|
|
|
2
|
|
|
3 <!-- [REQUIRED] Tool description displayed after the tool name -->
|
|
|
4 <description> </description>
|
|
|
5
|
|
|
6 <!-- [OPTIONAL] 3rd party tools, binaries, modules... required for the tool to work -->
|
|
|
7 <requirements>
|
|
|
8 <requirement type="binary">perl</requirement>
|
|
|
9 <requirement type="package" version="0.1.13">VCFtools</requirement>
|
|
|
10 </requirements>
|
|
|
11
|
|
|
12 <!-- [OPTIONAL] Command to be executed to get the tool's version string -->
|
|
|
13 <version_command>
|
|
|
14 <!--
|
|
|
15 tool_binary -v
|
|
|
16 -->
|
|
|
17 </version_command>
|
|
|
18
|
|
|
19 <!-- [REQUIRED] The command to execute -->
|
|
|
20 <command interpreter="perl">
|
|
|
21 vcfToolsStats.sh $filein $fileout_label $fileout_annot $fileout_het $fileout_imiss $fileout_sum $filelog
|
|
|
22 </command>
|
|
|
23
|
|
|
24 <!-- [REQUIRED] Input files and tool parameters -->
|
|
|
25 <inputs>
|
|
|
26 <param name="filein" type="data" format="vcf" optional="false" label="VCF input" />
|
|
|
27 <param name="fileout_label" type="text" value="vcf_stats" optional="false" label="Output file basename"/>
|
|
|
28 </inputs>
|
|
|
29
|
|
|
30 <!-- [REQUIRED] Output files -->
|
|
|
31 <outputs>
|
|
|
32 <data name="fileout_annot" format="txt" label="${fileout_label}.annotation" />
|
|
|
33 <data name="fileout_het" format="txt" label="${fileout_label}.het" />
|
|
|
34 <data name="fileout_imiss" format="txt" label="${fileout_label}.imiss" />
|
|
|
35 <data name="fileout_sum" format="txt" label="${fileout_label}.TsTv.summary" />
|
|
|
36 <data name="filelog" format="txt" label="${fileout_label}.log" />
|
|
|
37 </outputs>
|
|
|
38
|
|
|
39 <!-- [STRONGLY RECOMMANDED] Exit code rules -->
|
|
|
40 <stdio>
|
|
|
41 <!-- [HELP] If no exit code rule is defined, the tool will stop if anything is written to STDERR -->
|
|
|
42 <exit_code range="1:" level="fatal" />
|
|
|
43 </stdio>
|
|
|
44
|
|
|
45 <!-- [OPTIONAL] Tests to be run manually by the Galaxy admin -->
|
|
|
46 <tests>
|
|
|
47 <!-- [HELP] Test files have to be in the ~/test-data directory -->
|
|
|
48 <test>
|
|
|
49 <param name="filein" value="sample.vcf" />
|
|
|
50 <output name="fileout_annot" file="result.annotation" />
|
|
|
51 <output name="fileout_het" file="result.het" />
|
|
|
52 <output name="fileout_imiss" file="result.imiss" />
|
|
|
53 <output name="fileout_sum" file="result.TsTv.summary" />
|
|
|
54 <output name="filelog" file="result.log" />
|
|
|
55 </test>
|
|
|
56 </tests>
|
|
|
57
|
|
|
58 <!-- [OPTIONAL] Help displayed in Galaxy -->
|
|
|
59 <help>
|
|
|
60
|
|
|
61 .. class:: infomark
|
|
|
62
|
|
|
63 **Authors**
|
|
|
64
|
|
|
65 ---------------------------------------------------
|
|
|
66
|
|
|
67 .. class:: infomark
|
|
|
68
|
|
|
69 **Please cite** If you use this tool, please cite Dereeper et al. 2015 in prep.
|
|
|
70
|
|
|
71 ---------------------------------------------------
|
|
|
72
|
|
|
73 ================
|
|
|
74 VCF tools filter
|
|
|
75 ================
|
|
|
76
|
|
|
77 -----------
|
|
|
78 Description
|
|
|
79 -----------
|
|
|
80
|
|
|
81 Compute statistics on VCF file
|
|
|
82
|
|
|
83 -----------------
|
|
|
84 Workflow position
|
|
|
85 -----------------
|
|
|
86
|
|
|
87 **Upstream tools**
|
|
|
88
|
|
|
89 =========== ========================== =======
|
|
|
90 Name output file(s) format
|
|
|
91 =========== ========================== =======
|
|
|
92 =========== ========================== =======
|
|
|
93
|
|
|
94
|
|
|
95 **Downstream tools**
|
|
|
96
|
|
|
97 =========== ========================== =======
|
|
|
98 Name output file(s) format
|
|
|
99 =========== ========================== =======
|
|
|
100 =========== ========================== =======
|
|
|
101
|
|
|
102
|
|
|
103 ----------
|
|
|
104 Input file
|
|
|
105 ----------
|
|
|
106
|
|
|
107 VCF file
|
|
|
108 VCF file with all SNPs
|
|
|
109
|
|
|
110 ----------
|
|
|
111 Parameters
|
|
|
112 ----------
|
|
|
113
|
|
|
114 Output file basename
|
|
|
115 Prefix for the output VCF file
|
|
|
116
|
|
|
117 ------------
|
|
|
118 Output files
|
|
|
119 ------------
|
|
|
120
|
|
|
121 .annotation file
|
|
|
122 Statistics on annotation/location along genome
|
|
|
123
|
|
|
124 .het file
|
|
|
125 Statistics on heterozygosity of the individuals
|
|
|
126
|
|
|
127 .imiss
|
|
|
128 Statistics on missing data of the inidividuals
|
|
|
129 .TsTv.summary
|
|
|
130 Statistics on mutation types and transition/transvertion number
|
|
|
131
|
|
|
132 .log file
|
|
|
133
|
|
|
134 ---------------------------------------------------
|
|
|
135
|
|
|
136 ---------------
|
|
|
137 Working example
|
|
|
138 ---------------
|
|
|
139
|
|
|
140 Input files
|
|
|
141 ===========
|
|
|
142
|
|
|
143 VCF file
|
|
|
144 ---------
|
|
|
145
|
|
|
146 ::
|
|
|
147
|
|
|
148 #fileformat=VCFv4.1
|
|
|
149 #FILTER=<ID=LowQual,Description="Low quality">
|
|
|
150 #FORMAT=<ID=AD,Number=.,Type=Integer,Description="Allelic depths for the ref and alt alleles in the order listed">
|
|
|
151 [...]
|
|
|
152 CHROM POS ID REF ALT QUAL FILTER INFO FORMAT CATB1
|
|
|
153 chr1 2209 . G T 213.84 . AC=2;AF=1.00;AN=2;DP=7;Dels=0.00;FS=0.000;HaplotypeScore=0.0000;MLEAC=2;MLEAF=1.00;MQ=41.50;MQ0=0;QD=30.55;EFF=DOWNSTREAM(MODIFIER||||Cc01g00020|mRNA||GSCOCT00012438001|),UPSTREAM(MODIFIER||||Cc01g00010|mRNA||GSCOCT00012439001|) GT:AD:DP:GQ:PL 1/1:0,7:7:18:242,18,0
|
|
|
154
|
|
|
155
|
|
|
156 Parameters
|
|
|
157 ==========
|
|
|
158
|
|
|
159 Output name -> vcf_stat
|
|
|
160
|
|
|
161
|
|
|
162 Output files
|
|
|
163 ============
|
|
|
164
|
|
|
165 .annotation file
|
|
|
166 ----------------
|
|
|
167
|
|
|
168 ::
|
|
|
169
|
|
|
170 Genic 4489
|
|
|
171 Intergenic 466
|
|
|
172 ========
|
|
|
173 Intron 960
|
|
|
174 Exon 3248
|
|
|
175 UTR 281
|
|
|
176 ========
|
|
|
177 Non-syn 226
|
|
|
178 Synonym 3022
|
|
|
179
|
|
|
180 .het file
|
|
|
181 ---------
|
|
|
182
|
|
|
183 ::
|
|
|
184
|
|
|
185 INDV O(HOM) E(HOM) N_SITES F
|
|
|
186 CATB1 0 0.0 3616 0.00000
|
|
|
187
|
|
|
188 .imiss file
|
|
|
189 -----------
|
|
|
190
|
|
|
191 ::
|
|
|
192
|
|
|
193 INDV N_DATA N_GENOTYPES_FILTERED N_MISS F_MISS
|
|
|
194 CATB1 4813 0 0 0
|
|
|
195
|
|
|
196 .TsTv.summary file
|
|
|
197 ------------------
|
|
|
198
|
|
|
199 ::
|
|
|
200
|
|
|
201 MODEL COUNT
|
|
|
202 AC 371
|
|
|
203 AG 1467
|
|
|
204 AT 562
|
|
|
205 CG 330
|
|
|
206 CT 1659
|
|
|
207 GT 397
|
|
|
208 Ts 3126
|
|
|
209 Tv 1660
|
|
|
210
|
|
|
211
|
|
|
212 </help>
|
|
|
213
|
|
|
214 </tool>
|
