|
20
|
1 <tool id="cg_snpdiff" name="snpdiff 1.6" version="1.0.1">
|
|
|
2 <!--
|
|
|
3 This tool creates a GUI for the snpdiff function of cgatools from Complete Genomics, Inc.
|
|
|
4 written 6-18-2012 by bcrain@completegenomics.com
|
|
|
5 updated 8-13-2012 by bcrain@completegenomics.com
|
|
|
6 -->
|
|
|
7
|
|
|
8 <description>compares snp calls to var or masterVar file.</description> <!--adds description in toolbar-->
|
|
|
9
|
|
|
10 <command> <!--run executable-->
|
|
|
11 <!-- print version of cgatools to STDOUT-->
|
|
|
12 cgatools | head -1;
|
|
|
13
|
|
|
14 <!-- print command lines to STDOUT-->
|
|
|
15 echo "cgatools snpdiff
|
|
|
16 --reference $crr.fields.path
|
|
|
17 --variants $data_sources.varfile
|
|
|
18 --genotypes $genotype
|
|
|
19 --output-prefix cg_
|
|
|
20 --reports `echo ${report1} ${report2} ${report3} | sed 's/ */,/g'`
|
|
|
21 ";
|
|
|
22
|
|
|
23 <!-- execute cgatools-->
|
|
|
24 cgatools snpdiff
|
|
|
25 --reference $crr.fields.path
|
|
|
26 --variants $data_sources.varfile
|
|
|
27 --genotypes $genotype
|
|
|
28 --output-prefix cg_
|
|
|
29 --reports `echo ${report1} ${report2} ${report3} | sed 's/ */,/g'`
|
|
|
30
|
|
|
31 </command>
|
|
|
32
|
|
|
33 <outputs>
|
|
|
34 <data format="tabular" name="output1" from_work_dir="cg_Output.tsv" label="${tool.name} Output">
|
|
|
35 <filter>(report1 == 'Output')</filter>
|
|
|
36 </data>
|
|
|
37 <data format="tabular" name="output2" from_work_dir="cg_Verbose.tsv" label="${tool.name} Verbose">
|
|
|
38 <filter>(report2 == 'Verbose')</filter>
|
|
|
39 </data>
|
|
|
40 <data format="tabular" name="output3" from_work_dir="cg_Stats.tsv" label="${tool.name} Stats">
|
|
|
41 <filter>(report3 == 'Stats')</filter>
|
|
|
42 </data>
|
|
|
43 </outputs>
|
|
|
44
|
|
|
45 <inputs>
|
|
|
46 <!--form field to select crr file-->
|
|
|
47 <param name="crr" type="select" label="Reference genome (.crr file)">
|
|
|
48 <options from_data_table="cg_crr_files" />
|
|
|
49 </param>
|
|
|
50
|
|
|
51 <!--conditional to select data in/outside galaxy-->
|
|
|
52 <conditional name="data_sources">
|
|
|
53 <param name="data_source" type="select" label="Where is the input var or masterVar file?">
|
|
|
54 <option value="in" selected="true">imported into Galaxy</option>
|
|
|
55 <option value="out">located outside Galaxy (data on server or mounted drive)</option>
|
|
|
56 </param>
|
|
|
57
|
|
|
58 <!--form field to select input file-->
|
|
|
59 <when value="in">
|
|
|
60 <param name="varfile" type="data" format="cg_var,cg_mastervar" label="Var or masterVar file">
|
|
|
61 <validator type="dataset_ok_validator" />
|
|
|
62 <validator type="dataset_metadata_in_file" filename="cg_crr_files.loc"
|
|
|
63 metadata_name="dbkey" metadata_column="1"
|
|
|
64 message="cgatools is not currently available for this build."/>
|
|
|
65 </param>
|
|
|
66 </when>
|
|
|
67
|
|
|
68 <!--form field to enter input file-->
|
|
|
69 <when value="out">
|
|
|
70 <param name="varfile" type="text" label="Var or masterVar file (/path/file)" size="40" help="Var or masterVar file can be compressed (gz, bz2), e.g. /harddrive/GS00000XXXX-DID/GS00000YYYY-ASM/GS00123-DNA_G01_2000/ASM/var-GS00000YYYY-ASM.tsv.bz2">
|
|
|
71 <validator type="empty_field" message="You must supply a var or masterVar file"/>
|
|
|
72 </param>
|
|
|
73 </when>
|
|
|
74 </conditional>
|
|
|
75
|
|
|
76 <!--param to select genotypes file input-->
|
|
|
77 <param name="genotype" type="data" format="tabular" label="Genotypes file with SNP calls" help="The genotypes file is a tab-delimited file with at least the following columns: Chromosome, Offset0Based, Genotypes (Optional)">
|
|
|
78 <validator type="dataset_ok_validator" />
|
|
|
79 <validator type="dataset_metadata_in_file" filename="cg_crr_files.loc"
|
|
|
80 metadata_name="dbkey" metadata_column="1"
|
|
|
81 message="cgatools is not currently available for this build."/>
|
|
|
82 <!--<validator type="expression" message="Dataset does not match selected build.">value.dbkey == $crr.fields.dbkey</validator>-->
|
|
|
83 </param>
|
|
|
84
|
|
|
85 <!--params to select reports-->
|
|
|
86 <param name="report1" type="select" label="Create report Output">
|
|
|
87 <option value="">no</option>
|
|
|
88 <option value="Output" selected="true">yes</option>
|
|
|
89 </param>
|
|
|
90 <param name="report2" type="select" label="Create report Verbose">
|
|
|
91 <option value="">no</option>
|
|
|
92 <option value="Verbose">yes</option>
|
|
|
93 </param>
|
|
|
94 <param name="report3" type="select" label="Create report Stats">
|
|
|
95 <option value="">no</option>
|
|
|
96 <option value="Stats">yes</option>
|
|
|
97 </param>
|
|
|
98 </inputs>
|
|
|
99
|
|
|
100 <help>
|
|
|
101
|
|
|
102 **What it does**
|
|
|
103
|
|
|
104 This tool ompares snp calls to a Complete Genomics variant file.
|
|
|
105
|
|
|
106 **cgatools 1.6.0 Documentation**
|
|
|
107
|
|
|
108 Userguide: http://cgatools.sourceforge.net/docs/1.6.0/cgatools-user-guide.pdf
|
|
|
109
|
|
|
110 Release notes: http://cgatools.sourceforge.net/docs/1.6.0/cgatools-release-notes.pdf
|
|
|
111
|
|
|
112 **Command line reference**::
|
|
|
113
|
|
|
114 COMMAND NAME
|
|
|
115 snpdiff - Compares snp calls to a Complete Genomics variant file.
|
|
|
116
|
|
|
117 DESCRIPTION
|
|
|
118 Compares the snp calls in the "genotypes" file to the calls in a Complete
|
|
|
119 Genomics variant file. The genotypes file is a tab-delimited file with at
|
|
|
120 least the following columns (additional columns may be given):
|
|
|
121
|
|
|
122 Chromosome (Required) The name of the chromosome.
|
|
|
123 Offset0Based (Required) The 0-based offset in the chromosome.
|
|
|
124 GenotypesStrand (Optional) The strand of the calls in the Genotypes
|
|
|
125 column (+ or -, defaults to +).
|
|
|
126 Genotypes (Optional) The calls, one per allele. The following
|
|
|
127 calls are recognized:
|
|
|
128 A,C,G,T A called base.
|
|
|
129 N A no-call.
|
|
|
130 - A deleted base.
|
|
|
131 . A non-snp variation.
|
|
|
132
|
|
|
133 The output is a tab-delimited file consisting of the columns of the
|
|
|
134 original genotypes file, plus the following additional columns:
|
|
|
135
|
|
|
136 Reference The reference base at the given position.
|
|
|
137 VariantFile The calls made by the variant file, one per allele.
|
|
|
138 The character codes are the same as is described for
|
|
|
139 the Genotypes column.
|
|
|
140 DiscordantAlleles (Only if Genotypes is present) The number of
|
|
|
141 Genotypes alleles that are discordant with calls in
|
|
|
142 the VariantFile. If the VariantFile is described as
|
|
|
143 haploid at the given position but the Genotypes is
|
|
|
144 diploid, then each genotype allele is compared
|
|
|
145 against the haploid call of the VariantFile.
|
|
|
146 NoCallAlleles (Only if Genotypes is present) The number of
|
|
|
147 Genotypes alleles that were no-called by the
|
|
|
148 VariantFile. If the VariantFile is described as
|
|
|
149 haploid at the given position but the Genotypes is
|
|
|
150 diploid, then a VariantFile no-call is counted twice.
|
|
|
151
|
|
|
152 The verbose output is a tab-delimited file consisting of the columns of the
|
|
|
153 original genotypes file, plus the following additional columns:
|
|
|
154
|
|
|
155 Reference The reference base at the given position.
|
|
|
156 VariantFile The call made by the variant file for one allele (there is
|
|
|
157 a line in this file for each allele). The character codes
|
|
|
158 are the same as is described for the Genotypes column.
|
|
|
159 [CALLS] The rest of the columns are pasted in from the VariantFile,
|
|
|
160 describing the variant file line used to make the call.
|
|
|
161
|
|
|
162 The stats output is a comma-separated file with several tables describing
|
|
|
163 the results of the snp comparison, for each diploid genotype. The tables
|
|
|
164 all describe the comparison result (column headers) versus the genotype
|
|
|
165 classification (row labels) in different ways. The "Locus classification"
|
|
|
166 tables have the most detailed match classifications, while the "Locus
|
|
|
167 concordance" tables roll these match classifications up into "discordance"
|
|
|
168 and "no-call". A locus is considered discordant if it is discordant for
|
|
|
169 either allele. A locus is considered no-call if it is concordant for both
|
|
|
170 alleles but has a no-call on either allele. The "Allele concordance"
|
|
|
171 describes the comparison result on a per-allele basis.
|
|
|
172
|
|
|
173 OPTIONS
|
|
|
174 -h [ --help ]
|
|
|
175 Print this help message.
|
|
|
176
|
|
|
177 --reference arg
|
|
|
178 The input crr file.
|
|
|
179
|
|
|
180 --variants arg
|
|
|
181 The input variant file.
|
|
|
182
|
|
|
183 --genotypes arg
|
|
|
184 The input genotypes file.
|
|
|
185
|
|
|
186 --output-prefix arg
|
|
|
187 The path prefix for all output reports.
|
|
|
188
|
|
|
189 --reports arg (=Output,Verbose,Stats)
|
|
|
190 Comma-separated list of reports to generate. A report is one of:
|
|
|
191 Output The output genotypes file.
|
|
|
192 Verbose The verbose output file.
|
|
|
193 Stats The stats output file.
|
|
|
194
|
|
|
195 SUPPORTED FORMAT_VERSION
|
|
|
196 0.3 or later
|
|
|
197 </help>
|
|
|
198 </tool>
|
