|
0
|
1 <tool id="cg_evidence2sam" name="evidence2sam(beta) 1.6" version="1.0.0">
|
|
|
2 <!--
|
|
|
3 This tool creates a GUI for the evidence2sam function of cgatools from Complete Genomics, Inc.
|
|
|
4 written 8-31-2012 by bcrain@completegenomics.com
|
|
|
5 -->
|
|
|
6
|
|
|
7 <description>converts evidence mappings to SAM format</description>
|
|
|
8
|
|
|
9 <command>
|
|
|
10 <!-- print version of cgatools to STDOUT-->
|
|
|
11 cgatools | head -1;
|
|
|
12
|
|
|
13 <!-- print command lines to STDOUT-->
|
|
|
14 echo "cgatools evidence2sam --beta
|
|
|
15 --reference $crr.fields.path
|
|
|
16 --output $output
|
|
|
17 --evidence-dnbs $data_sources.input
|
|
|
18 --consistent-mapping-range $range
|
|
|
19 #if $region.selectregion == "yes"
|
|
|
20 --extract-genomic-region $region.coordinates
|
|
|
21 #end if
|
|
|
22 $duplicates
|
|
|
23 $mates
|
|
|
24 $intervals
|
|
|
25 $skip
|
|
|
26 $svcandidates
|
|
|
27 $unmapped
|
|
|
28 $primary
|
|
|
29 ";
|
|
|
30
|
|
|
31 <!-- execute cgatools-->
|
|
|
32 cgatools evidence2sam --beta
|
|
|
33 --reference $crr.fields.path
|
|
|
34 --evidence-dnbs $data_sources.input
|
|
|
35 #if $region.selectregion == "yes"
|
|
|
36 --extract-genomic-region $region.coordinates
|
|
|
37 #end if
|
|
|
38 $duplicates
|
|
|
39 $mates
|
|
|
40 $intervals
|
|
|
41 $skip
|
|
|
42 $svcandidates
|
|
|
43 $unmapped
|
|
|
44 $primary
|
|
|
45 --consistent-mapping-range $range
|
|
|
46 --output $output
|
|
|
47 </command>
|
|
|
48
|
|
|
49 <outputs>
|
|
|
50 <data format="tabular" name="output" label="${tool.name} output"/>
|
|
|
51 </outputs>
|
|
|
52
|
|
|
53 <inputs>
|
|
|
54 <!--form field to select crr file-->
|
|
|
55 <param name="crr" type="select" label="Reference genome (.crr file)">
|
|
|
56 <options from_data_table="cg_crr_files" />
|
|
|
57 </param>
|
|
|
58
|
|
|
59 <!--conditional to select input file-->
|
|
|
60 <conditional name="data_sources">
|
|
|
61 <param name="data_source" type="select" label="Where is the input evidence file?">
|
|
|
62 <option value="in">imported into Galaxy</option>
|
|
|
63 <option value="out" selected="true">located outside Galaxy (data on server or mounted drive)</option>
|
|
|
64 </param>
|
|
|
65
|
|
|
66 <!--form field to select evidence files-->
|
|
|
67 <when value="in">
|
|
|
68 <param name="input" type="data" format="tabluar" label="EvidenceDnbs file">
|
|
|
69 <validator type="dataset_ok_validator" />
|
|
|
70 <validator type="dataset_metadata_in_file" filename="cg_crr_files.loc"
|
|
|
71 metadata_name="dbkey" metadata_column="1"
|
|
|
72 message="cgatools is not currently available for this build."/>
|
|
|
73 </param>
|
|
|
74 </when>
|
|
|
75
|
|
|
76 <!--form field to enter external input file-->
|
|
|
77 <when value="out">
|
|
|
78 <param name="input" type="text" label="EvidenceDnbs file (/path/file)" size="40" help="e.g. /harddrive/GS00000XXXX-DID/GS00000YYYY-ASM/GS00123-DNA_G01_2000/ASM/EVIDENCE/evidenceDnbs-chr21-GS00000YYYY-ASM.tsv.bz2">
|
|
|
79 <validator type="empty_field" message="You must supply an evidenceDnbs file"/>
|
|
|
80 </param>
|
|
|
81 </when>
|
|
|
82 </conditional>
|
|
|
83
|
|
|
84 <!--form field to select chromosomal region-->
|
|
|
85 <conditional name="region">
|
|
|
86 <param name="selectregion" type="select" label="Do you what to extract specific genomic region?">
|
|
|
87 <option value="no" selected="true">no</option>
|
|
|
88 <option value="yes">yes</option>
|
|
|
89 </param>
|
|
|
90
|
|
|
91 <when value="yes">
|
|
|
92 <param name="coordinates" type="text" label="Enter genomic coordinates to avoid converting the entire file (chr,from,to)" size="40" help="Specify the region as a half-open interval chr,from,to (e.g. chrX,15203639,15412498)"/>
|
|
|
93 </when>
|
|
|
94 </conditional>
|
|
|
95
|
|
|
96 <!--form field to select duplicate handling-->
|
|
|
97 <param name="duplicates" type="select" label="Keep local duplicates of DNB mappings (default no)" help="All the output SAM records will be marked as 'not primary' if this option is used.">
|
|
|
98 <option value="" selected="true">no</option>
|
|
|
99 <option value="--keep-duplicates">yes</option>
|
|
|
100 </param>
|
|
|
101
|
|
|
102 <!--form field to generate mate sequence-->
|
|
|
103 <param name="mates" type="select" label="Generate mate sequence (R2) and score (Q2) tags (default no)">
|
|
|
104 <option value="" selected="true">no</option>
|
|
|
105 <option value="--add-mate-sequence">yes</option>
|
|
|
106 </param>
|
|
|
107
|
|
|
108 <!--form field to generate interval ids-->
|
|
|
109 <param name="intervals" type="select" label="Generate interval id (ZI:I) and allele id (ZA:I) tags (default no)">
|
|
|
110 <option value="" selected="true">no</option>
|
|
|
111 <option value="--add-allele-id">yes</option>
|
|
|
112 </param>
|
|
|
113
|
|
|
114 <!--form field to skip not mapped reads-->
|
|
|
115 <param name="skip" type="select" label="Skip not mapped records (default no)">
|
|
|
116 <option value="" selected="true">no</option>
|
|
|
117 <option value="--skip-not-mapped">yes</option>
|
|
|
118 </param>
|
|
|
119
|
|
|
120 <!--form field to skip not mapped reads-->
|
|
|
121 <param name="svcandidates" type="select" label="Mate unique single arm mappings in SAM including those on different stands and chromosomes (default no)">
|
|
|
122 <option value="" selected="true">no</option>
|
|
|
123 <option value="--mate-sv-candidates">yes</option>
|
|
|
124 </param>
|
|
|
125
|
|
|
126 <!--form field to skip not mapped reads-->
|
|
|
127 <param name="unmapped" type="select" label="Generate mate sequence and score tags for inconsistent mappings only (default no)">
|
|
|
128 <option value="" selected="true">no</option>
|
|
|
129 <option value="--add-unmapped-mate-info">yes</option>
|
|
|
130 </param>
|
|
|
131
|
|
|
132 <!--form field to skip not mapped reads-->
|
|
|
133 <param name="primary" type="select" label="Use primary mappings only (default no)" help="Report only the best mappings">
|
|
|
134 <option value="" selected="true">no</option>
|
|
|
135 <option value="--primary-mappings-only">yes</option>
|
|
|
136 </param>
|
|
|
137
|
|
|
138 <param name="range" type="integer" value="1300" label="Maximum distance between consistent mates (default 1300)">
|
|
|
139 <validator type="empty_field" message="You must enter a value, the default is 1300" />
|
|
|
140 </param>
|
|
|
141 </inputs>
|
|
|
142
|
|
|
143 <stdio>
|
|
|
144 <regex match="failed" source="stderr" level="fatal"/>
|
|
|
145 <regex match="error" source="stderr" level="fatal"/>
|
|
|
146 <regex match="Export the sequence:" source="stderr" level="warning" description="Finished:" />
|
|
|
147 </stdio>
|
|
|
148
|
|
|
149 <help>
|
|
|
150
|
|
|
151 **What it does**
|
|
|
152
|
|
|
153 This tool uses cgatools evidence2sam to convert Complete Genomics evidence mappings to SAM format
|
|
|
154
|
|
|
155 **cgatools 1.6.0 Documentation**
|
|
|
156
|
|
|
157 Userguide: http://cgatools.sourceforge.net/docs/1.6.0/cgatools-user-guide.pdf
|
|
|
158
|
|
|
159 Release notes: http://cgatools.sourceforge.net/docs/1.6.0/cgatools-release-notes.pdf
|
|
|
160
|
|
|
161 **Command line reference**::
|
|
|
162
|
|
|
163 COMMAND NAME
|
|
|
164 evidence2sam - Converts CGI variant evidence data into SAM format.
|
|
|
165
|
|
|
166 DESCRIPTION
|
|
|
167 The evidence2sam converter takes as input evidence mapping files
|
|
|
168 (evidenceDnbs-*) and generates one SAM file as an output. The output is
|
|
|
169 sent into stdout by default. By default, all the evidence mapping records
|
|
|
170 from the input are converted into a pair of corresponding SAM records - one
|
|
|
171 record for each HalfDNB. The negative gaps in CGI mappings are represented
|
|
|
172 using GS/GQ/GC tags.
|
|
|
173
|
|
|
174 OPTIONS
|
|
|
175 -h [ --help ]
|
|
|
176 Print this help message.
|
|
|
177
|
|
|
178 --beta
|
|
|
179 This is a beta command. To run this command, you must pass the --beta
|
|
|
180 flag.
|
|
|
181
|
|
|
182 -e [ --evidence-dnbs ] arg
|
|
|
183 Input evidence dnbs file.
|
|
|
184
|
|
|
185 -s [ --reference ] arg
|
|
|
186 Reference file.
|
|
|
187
|
|
|
188 -o [ --output ] arg (=STDOUT)
|
|
|
189 The output SAM file (may be omitted for stdout).
|
|
|
190
|
|
|
191 -r [ --extract-genomic-region ] arg
|
|
|
192 defines a region as a half-open interval 'chr,from,to'.
|
|
|
193
|
|
|
194 --keep-duplicates
|
|
|
195 Keep local duplicates of DNB mappings.All the output SAM records will
|
|
|
196 be marked as not primary if this option is used.
|
|
|
197
|
|
|
198 --add-allele-id
|
|
|
199 Generate interval id and allele id tags.
|
|
|
200
|
|
|
201 --skip-not-mapped
|
|
|
202 Skip not mapped records
|
|
|
203
|
|
|
204 --add-mate-sequence
|
|
|
205 Generate mate sequence and score tags.
|
|
|
206
|
|
|
207 --mate-sv-candidates
|
|
|
208 Inconsistent mappings are normally converted as single arm mappings
|
|
|
209 with no mate information provided. If the option is used map2sam will
|
|
|
210 mate unique single arm mappings in SAM including those on different
|
|
|
211 stands and chromosomes. To distinguish these "artificially" mated
|
|
|
212 records a tag "XS:i:1" is used. The MAPQ provided for these records is
|
|
|
213 a single arm mapping weight.
|
|
|
214
|
|
|
215 --add-unmapped-mate-info
|
|
|
216 works like add-mate-sequence, but is applied to inconsistent mappings
|
|
|
217 only
|
|
|
218
|
|
|
219 --primary-mappings-only
|
|
|
220 report only the best mappings
|
|
|
221
|
|
|
222 --consistent-mapping-range arg (=1300)
|
|
|
223 limit the maximum distance between consistent mates
|
|
|
224
|
|
|
225
|
|
|
226 SUPPORTED FORMAT_VERSION
|
|
|
227 0.3 or later
|
|
|
228 </help>
|
|
|
229 </tool>
|
