comparison SVDetect_run_parallel.xml @ 4:f7a84d31bd83 draft

Uploaded
author bzeitouni
date Mon, 11 Jun 2012 12:30:38 -0400
parents
children
comparison
equal deleted inserted replaced
3:861783bb65d2 4:f7a84d31bd83
1 <tool id="svdetect_run_parallel" name="Detect clusters of anomalously mapped pairs">
2
3 <description>and identify structural variants</description>
4
5 <command interpreter="perl">SVDetect_run_parallel.pl
6
7 #if $getLinks.linking == "linking"
8 linking
9 <!-- -out1 '$links_file' -->
10 #end if
11 #if $getFilteredLinks.filtering == "filtering"
12 filtering
13 <!--- out2 '$flinks_file' -->
14 #if str($getFilteredLinks.links2SV) == "create"
15 links2SV
16 -out3 '$sv_file'
17 #end if
18 #if $getFilteredLinks.file_conversion.file_conversion_select=="convert" and str($getFilteredLinks.file_conversion.links2circos) == "create"
19 links2circos
20 -out4 '$circos_file'
21 #end if
22 #if $getFilteredLinks.file_conversion.file_conversion_select=="convert" and str($getFilteredLinks.file_conversion.links2bed) == "create"
23 links2bed
24 -out5 '$bed_file'
25 #end if
26 #end if
27 -conf '$config_file'
28 -l '$log_file'
29 -N '$sample_name'
30
31 </command>
32
33 <inputs>
34 <param name="sample_name" type="text" value="sample" label="Sample Name"/>
35 <param name="mates_file" format="bam" type="data" label="Input BAM file (.ab.bam)"/>
36 <param name="cmap_file" format="len" type="data" label="Chromosomes list file (.len)" help="Tabulated file format with Chromosome ID (integer from 1), name and length"/>
37 <param name="mates_orientation" type="select" format="txt" label="Type of sequencing technology and libraries">
38 <option value="FR">Illumina paired-ends</option>
39 <option value="RF">Illumina mate-pairs</option>
40 <option value="FR">SOLiD paired-ends</option>
41 <option value="RR">SOLiD mate-pairs</option>
42 </param>
43 <param name="read1_length" type="integer" size="10" value="50" label="Read 1 length (bp)" help="Length of the first read in a pair (left read)"/>
44 <param name="read2_length" type="integer" size="10" value="50" label="Read 2 length (bp)" help="Length of the second read in a pair (right read)"/>
45 <param name="sv_type" type="select" format="txt" label="Type of SV to detect">
46 <option value="all">all types of SVs</option>
47 <option value="intra">intrachromosomal SVs only</option>
48 <option value="inter">interchromosomal SVs only</option>
49 </param>
50
51 <conditional name="getLinks">
52 <param name="linking" type="select" label="Linking procedure" help="Detection and isolation of links">
53 <option value="linking">Yes</option>
54 <option value="">No, already done</option>
55 </param>
56 <when value="">
57 <!-- do nothing here -->
58 </when>
59 <when value="linking">
60 <param name="splitmate" label="Do you want to split the original mate file per chromosome for parallel computing?" type="boolean" truevalue="split" falsevalue="do_not_split" checked="True" help="Untick it if already done"/>
61 <param name="window_size" type="integer" size="20" value="3000" label="Window size (bp)" help="Equal to at least “2µ+2√2σ"/>
62 <param name="step_length" type="integer" size="20" value="250" label="Step length size (bp)" help="Equal to 1/2 or 1/4 of the window size"/>
63 </when>
64 </conditional>
65
66 <conditional name="getFilteredLinks">
67 <param name="filtering" type="select" label="Filtering procedure" help="Filtering of links according different parameters and thresholds">
68 <option value="filtering">Yes</option>
69 <option value="">No</option>
70 </param>
71 <when value="">
72 <!-- do nothing here -->
73 </when>
74 <when value="filtering">
75
76 <param name="splitlink" label="Do you want to split the original link file per chromosome for parallel computing?" type="boolean" truevalue="split" falsevalue="do_not_split" checked="False" help="Untick it if (the linking is) already done"/>
77 <param name="chromosomes" type="text" size="20" label="List of chromosome names to keep or exclude"/>
78 <param name="nb_pairs_threshold" type="integer" size="20" value="5" label="Minimum number of pairs in a cluster"/>
79
80 <conditional name="filter1">
81 <param name="strand_filtering" type="select" label="Strand filtering procedure">
82 <option value="strand">Yes</option>
83 <option value="">No</option>
84 </param>
85 <when value="">
86 <!-- do nothing here -->
87 </when>
88 <when value="strand">
89
90 <conditional name="filter2">
91 <param name="order_filtering" type="select" label="Order filtering procedure">
92 <option value="order">Yes</option>
93 <option value="">No</option>
94 </param>
95 <when value="">
96 <!-- do nothing here -->
97 </when>
98 <when value="order">
99
100 <conditional name="filter3">
101 <param name="insert_size_filtering" type="select" label="Insert-size filtering procedure">
102 <option value="insert">Yes</option>
103 <option value="">No</option>
104 </param>
105 <when value="">
106 <!-- do nothing here -->
107 </when>
108 <when value="insert">
109 <param name="indel_sigma_threshold" type="float" size="20" value="3" label="Minimal number of sigma fold for the insert size filtering and to call insertions and deletions"/>
110 <param name="dup_sigma_threshold" type="float" size="20" value="3" label="minimal number of sigma fold for the insert size filtering to call tandem duplications"/>
111 <param name="singleton_sigma_threshold" type="float" size="20" value="4" label="Minimal number of sigma fold for the insert size filtering to call singletons" help="for Illumina mate-pairs only"/>
112 </when>
113 </conditional>
114
115 <param name="mu_length" type="integer" size="20" value="3000" label="Mean insert size value (µ) of normally mapped mate-pairs, in bp"/>
116 <param name="sigma_length" type="integer" size="20" value="250" label="Calculated sd value (σ) from the distribution of normally mapped mate-pairs, in bp"/>
117 <param name="nb_pairs_order_threshold" type="integer" size="20" value="2" label="Minimal number of pairs in a subgroup of paired-end reads for balanced events"/>
118 </when>
119 </conditional>
120
121 <param name="final_score_threshold" type="float" size="20" value="1.0" label="Minimal final filtering score for calling SVs" help="A value of 1 means all the pairs in a cluster were consistent between each other after applying filters"/>
122 </when>
123 </conditional>
124
125 <param name="links2SV" label="Do you want to have filtered links in a tabulated file format showing significant SVs?" type="boolean" truevalue="create" falsevalue="do_not_create" checked="True"/>
126
127 <conditional name="file_conversion">
128 <param name="file_conversion_select" type="select" label="Output file conversion" help="Converts filtered links to Circos/BED files format for graphical view of SVs">
129 <option value="do_not_convert">No</option>
130 <option value="convert">Yes</option>
131 </param>
132 <when value="do_not_convert">
133 <!-- do nothing here -->
134 </when>
135 <when value="convert">
136 <param name="links2circos" label="Converts the link list to the Circos link format" type="boolean" truevalue="create" falsevalue="do_not_create" checked="True"/>
137 <param name="links2bed" label="Converts the link list to the UCSC BED format" type="boolean" truevalue="create" falsevalue="do_not_create" checked="False"/>
138 <param name="organism_id" type="text" size="10" value="hs" label="Organism ID"/>
139 <repeat name="color_code" title="Color-code" min="1" max="7">
140 <param name="color" type="select" label="Color">
141 <option value="grey">grey</option>
142 <option value="black">black</option>
143 <option value="blue">blue</option>
144 <option value="green">green</option>
145 <option value="purple">purple</option>
146 <option value="orange">orange</option>
147 <option value="red">red</option>
148 </param>
149 <param name="interval" type="text" value="1,3" label="Interval"/>
150 </repeat>
151 </when>
152 </conditional>
153 </when>
154 </conditional>
155 </inputs>
156
157
158 <outputs>
159 <!--<data format="txt" name="links_file" label="svdetect.links">
160 <filter>getLinks['linking']=="linking"</filter>
161 </data>
162 <data format="txt" name="flinks_file" label="svdetect.links.filtered">
163 <filter>getFilteredLinks['filtering']=="filtering"</filter>
164 </data>-->
165 <data format="sv" name="sv_file" label="${sample_name}.sv">
166 <filter>(
167 getFilteredLinks['filtering']=="filtering" and
168 getFilteredLinks['links2SV'] is True
169 )
170 </filter>
171 </data>
172 <data format="segdup" name="circos_file" label="${sample_name}.segdup">
173 <filter>(
174 getFilteredLinks['filtering']=="filtering" and
175 getFilteredLinks['file_conversion']['file_conversion_select']=="convert" and
176 getFilteredLinks['file_conversion']['links2circos'] is True
177 )
178 </filter>
179 </data>
180 <data format="bed" name="bed_file" label="${sample_name}.bed">
181 <filter>(
182 getFilteredLinks['filtering']=="filtering" and
183 getFilteredLinks['file_conversion']['file_conversion_select']=="convert" and
184 getFilteredLinks['file_conversion']['links2bed'] is True
185 )
186 </filter>
187 </data>
188 <data format="txt" name="log_file" label="${sample_name}.svdetect_run.log"/>
189 </outputs>
190
191
192
193 <configfiles>
194 <configfile name="config_file">
195 &lt;general&gt;
196 input_format = bam
197 sv_type = ${sv_type}
198 mates_orientation=${mates_orientation}
199 read1_length=${read1_length}
200 read2_length=${read2_length}
201 mates_file=${mates_file}
202 cmap_file=${cmap_file}
203 tmp_dir=$__new_file_path__/svdetect/tmp
204 output_dir=$__new_file_path__/svdetect
205 num_threads=8
206 &lt;/general&gt;
207
208 #if $getLinks.linking == "linking"
209 &lt;detection&gt;
210 #if str($getLinks.splitmate) == "split"
211 split_mate_file=1
212 #else
213 split_mate_file=0
214 #end if
215 window_size=${getLinks.window_size}
216 step_length=${getLinks.step_length}
217 &lt;/detection&gt;
218 #end if
219
220 #if $getFilteredLinks.filtering == "filtering"
221 &lt;filtering&gt;
222 #if str($getFilteredLinks.splitlink) == "split"
223 split_link_file=1
224 #else
225 split_link_file=0
226 #end if
227 #if str($getFilteredLinks.chromosomes) != ""
228 chromosomes=${getFilteredLinks.chromosomes}
229 #end if
230 nb_pairs_threshold=${getFilteredLinks.nb_pairs_threshold}
231 #if $getFilteredLinks.filter1.strand_filtering == "strand"
232 strand_filtering=1
233 final_score_threshold=${getFilteredLinks.filter1.final_score_threshold}
234 #if $getFilteredLinks.filter1.filter2.order_filtering == "order"
235 order_filtering=1
236 mu_length=${getFilteredLinks.filter1.filter2.mu_length}
237 sigma_length=${getFilteredLinks.filter1.filter2.sigma_length}
238 nb_pairs_order_threshold=${getFilteredLinks.filter1.filter2.nb_pairs_order_threshold}
239 #if $getFilteredLinks.filter1.filter2.filter3.insert_size_filtering == "insert"
240 insert_size_filtering=1
241 indel_sigma_threshold=${getFilteredLinks.filter1.filter2.filter3.indel_sigma_threshold}
242 dup_sigma_threshold=${getFilteredLinks.filter1.filter2.filter3.dup_sigma_threshold}
243 singleton_sigma_threshold=${getFilteredLinks.filter1.filter2.filter3.singleton_sigma_threshold}
244 #else
245 insert_size_filtering=0
246 #end if
247 #else
248 order_filtering=0
249 #end if
250 #else
251 strand_filtering=0
252 #end if
253 &lt;/filtering&gt;
254 #end if
255
256 #if $getFilteredLinks.filtering == "filtering"
257 #if $getFilteredLinks.file_conversion.file_conversion_select == "convert"
258 #if str($getFilteredLinks.file_conversion.links2circos) == "create"
259 &lt;circos&gt;
260 organism_id=${getFilteredLinks.file_conversion.organism_id}
261 &lt;colorcode&gt;
262 #for $color_repeat in $getFilteredLinks.file_conversion.color_code
263 ${color_repeat.color}=${color_repeat.interval}
264 #end for
265 &lt;/colorcode&gt;
266 &lt;/circos&gt;
267 #end if
268 #if str($getFilteredLinks.file_conversion.links2bed) == "create"
269 &lt;bed&gt;
270 &lt;colorcode&gt;
271 #for $color_repeat in $getFilteredLinks.file_conversion.color_code
272 #if str($color_repeat.color)== "grey"
273 190,190,190=${color_repeat.interval}
274 #end if
275 #if str($color_repeat.color)== "black"
276 0,0,0=${color_repeat.interval}
277 #end if
278 #if str($color_repeat.color)== "blue"
279 0,0,255=${color_repeat.interval}
280 #end if
281 #if str($color_repeat.color)== "green"
282 0,255,0=${color_repeat.interval}
283 #end if
284 #if str($color_repeat.color)== "purple"
285 153,50,205=${color_repeat.interval}
286 #end if
287 #if str($color_repeat.color)== "orange"
288 255,140,0=${color_repeat.interval}
289 #end if
290 #if str($color_repeat.color)== "red"
291 255,0,0=${color_repeat.interval}
292 #end if
293 #end for
294 &lt;/colorcode&gt;
295 &lt;/bed&gt;
296 #end if
297 #end if
298 #end if
299 </configfile>
300 </configfiles>
301
302 <help>
303 **What it does**
304
305 SVDetect - Version : 0.8
306
307 Parallel version (nCPU=8)
308
309 SVDetect is a application for the isolation and the type prediction of intra- and inter-chromosomal rearrangements from paired-end/mate-pair sequencing data provided by the high-throughput sequencing technologies
310
311 This tool aims to identifying structural variations (SVs) with both clustering and sliding-window strategies, and helping in their visualization at the genome scale.
312 SVDetect is compatible with SOLiD and Illumina (>=1.3) reads.
313
314 Manual documentation available at the http://svdetect.sourceforge.net/Site/Manual.html
315
316 -----
317
318 .. class:: infomark
319
320 Contact Bruno Zeitouni (bruno.zeitouni@curie.fr) for any questions or concerns about the Galaxy implementation of SVDetect.
321
322 </help>
323
324 </tool>