|
0
|
1 <tool id="vsearch_search" name="VSearch search" version="@VERSION@.0">
|
|
|
2 <description></description>
|
|
|
3 <macros>
|
|
|
4 <import>vsearch_macros.xml</import>
|
|
|
5 </macros>
|
|
|
6 <expand macro="requirements" />
|
|
|
7 <expand macro="stdio" />
|
|
|
8 <expand macro="version_command" />
|
|
|
9 <command>
|
|
|
10 <![CDATA[
|
|
|
11 vsearch
|
|
|
12
|
|
|
13 @GENERAL@
|
|
|
14 --db $dbfile
|
|
|
15 --dbmask $dbmask
|
|
|
16
|
|
|
17 ##--fulldp full dynamic programming alignment (always on)
|
|
|
18 ##--gapext STRING penalties for gap extension (2I/1E)
|
|
|
19 ##--gapopen STRING penalties for gap opening (20I/2E)
|
|
|
20 $hardmask
|
|
|
21 --id $id
|
|
|
22 --iddef $iddef
|
|
|
23
|
|
|
24 #if $qmask != 'no':
|
|
|
25 --qmask $qmask
|
|
|
26 #end if
|
|
|
27
|
|
|
28 ## --rowlen INT width of alignment lines in alnout output (64)
|
|
|
29
|
|
|
30 $self_param
|
|
|
31 $selfid_param
|
|
|
32 #if $sizeout:
|
|
|
33 --sizeout $sizeout
|
|
|
34 #end if
|
|
|
35 --strand $strand
|
|
|
36 --usearch_global $queryfile
|
|
|
37
|
|
|
38 #if '--alnout' in str($outputs):
|
|
|
39 --alnout $alnout
|
|
|
40 #end if
|
|
|
41 #if '--blast6out' in str($outputs):
|
|
|
42 --blast6out $blast6out
|
|
|
43 #end if
|
|
|
44 #if '--dbmatched' in str($outputs):
|
|
|
45 --dbmatched $dbmatched
|
|
|
46 #end if
|
|
|
47 #if '--dbnotmatched' in str($outputs):
|
|
|
48 --dbnotmatched $dbnotmatched
|
|
|
49 #end if
|
|
|
50 #if '--fastapairs' in str($outputs):
|
|
|
51 --fastapairs $fastapairs
|
|
|
52 #end if
|
|
|
53 #if '--notmatched' in str($outputs):
|
|
|
54 --notmatched $notmatched
|
|
|
55 #end if
|
|
|
56 #if '--matched' in str($outputs):
|
|
|
57 --matched $matched
|
|
|
58 #end if
|
|
|
59
|
|
|
60 #if $adv_opts.adv_opts_selector == "advanced":
|
|
|
61 $adv_opts.top_hits_only
|
|
|
62 $adv_opts.rightjust
|
|
|
63 $adv_opts.leftjust
|
|
|
64 --target_cov $adv_opts.target_cov
|
|
|
65 --query_cov $adv_opts.query_cov
|
|
|
66
|
|
|
67
|
|
|
68 #if $adv_opts.maxid:
|
|
|
69 --maxid $adv_opts.maxid
|
|
|
70 #end if
|
|
|
71 #if $adv_opts.maxqt:
|
|
|
72 --maxqt $adv_opts.maxqsize
|
|
|
73 #end if
|
|
|
74 #if $adv_opts.maxsizeratio:
|
|
|
75 --maxsizeratio $adv_opts.maxsizeratio
|
|
|
76 #end if
|
|
|
77 #if $adv_opts.maxsl:
|
|
|
78 --maxsl $adv_opts.maxsl
|
|
|
79 #end if
|
|
|
80 #if $adv_opts.mid:
|
|
|
81 --mid $adv_opts.mid
|
|
|
82 #end if
|
|
|
83 #if $adv_opts.minqt:
|
|
|
84 --minqt $adv_opts.minqt
|
|
|
85 #end if
|
|
|
86 #if $adv_opts.minsizeratio:
|
|
|
87 --minsizeratio $adv_opts.minseqlength
|
|
|
88 #end if
|
|
|
89 #if $adv_opts.minsl:
|
|
|
90 --minsl $adv_opts.minsl
|
|
|
91 #end if
|
|
|
92
|
|
|
93 #if $adv_opts.mintsize:
|
|
|
94 --mintsize $adv_opts.mintsize
|
|
|
95 #end if
|
|
|
96 #if $adv_opts.mismatch:
|
|
|
97 --mismatch $adv_opts.mismatch
|
|
|
98 #end if
|
|
|
99 #if $adv_opts.maxqsize:
|
|
|
100 --maxqsize $adv_opts.maxqsize
|
|
|
101 #end if
|
|
|
102 #if $adv_opts.mincols:
|
|
|
103 --mincols $adv_opts.mincols
|
|
|
104 #end if
|
|
|
105 #if $adv_opts.maxsubs:
|
|
|
106 --maxsubs $adv_opts.maxsubs
|
|
|
107 #end if
|
|
|
108 #if $adv_opts.maxrejects:
|
|
|
109 --maxrejects $adv_opts.maxrejects
|
|
|
110 #end if
|
|
|
111 #if $adv_opts.maxaccepts:
|
|
|
112 --maxaccepts $adv_opts.maxaccepts
|
|
|
113 #end if
|
|
|
114 #if $adv_opts.maxdiffs:
|
|
|
115 --maxdiffs $adv_opts.maxdiffs
|
|
|
116 #end if
|
|
|
117 #if $adv_opts.maxgaps:
|
|
|
118 --maxgaps $adv_opts.maxgaps
|
|
|
119 #end if
|
|
|
120 #if $adv_opts.maxhits:
|
|
|
121 --maxhits $adv_opts.maxhits
|
|
|
122 #end if
|
|
|
123 #if $adv_opts.match:
|
|
|
124 --match $adv_opts.match
|
|
|
125 #end if
|
|
|
126 #if $adv_opts.idprefix:
|
|
|
127 --idprefix $adv_opts.idprefix
|
|
|
128 #end if
|
|
|
129 #if $adv_opts.idsuffix:
|
|
|
130 --idsuffix $adv_opts.idsuffix
|
|
|
131 #end if
|
|
|
132
|
|
|
133 #if $adv_opts.uclust_output.uclust_output_select == 'yes':
|
|
|
134 --uc
|
|
|
135 $adv_opts.uclust_output.uc_allhits
|
|
|
136 #end if
|
|
|
137
|
|
|
138 #if $adv_opts.userfields_output.userfields_output_select == 'yes':
|
|
|
139 --userfields '#echo '+'.join( str($adv_opts.userfields_output.userfields).split(',') )#'
|
|
|
140 --userout $userout
|
|
|
141 #end if
|
|
|
142
|
|
|
143 ##--weak_id REAL include aligned hits with >= id; continue search
|
|
|
144 --wordlength $adv_opts.wordlength
|
|
|
145
|
|
|
146 $$adv_opts.output_no_hits
|
|
|
147
|
|
|
148 #end if
|
|
|
149
|
|
|
150
|
|
|
151 ]]>
|
|
|
152 </command>
|
|
|
153 <inputs>
|
|
|
154 <param name="queryfile" type="data" format="fasta" label="Select your query FASTA file" help="(--usearch_global)" />
|
|
|
155 <param name="dbfile" type="data" format="fasta" label="Select your database FASTA file" help="(--db)" />
|
|
|
156
|
|
|
157 <expand macro="qmask" />
|
|
|
158 <expand macro="sizeout" />
|
|
|
159 <expand macro="strand" />
|
|
|
160 <expand macro="self_and_selfid" />
|
|
|
161 <expand macro="id_and_iddef" />
|
|
|
162 <expand macro="hardmask" />
|
|
|
163
|
|
|
164 <param name="dbmask" type="select" optional="True" label="Mask database sequences" help="(--dbmask)">
|
|
|
165 <option value="none" selected="True">No masking</option>
|
|
|
166 <option value="dust" selected="True">dust</option>
|
|
|
167 <option value="soft">soft</option>
|
|
|
168 </param>
|
|
|
169
|
|
|
170 <expand macro="general_output">
|
|
|
171 <option value="--dbmatched">FASTA file for matching database sequences</option>
|
|
|
172 <option value="--dbnotmatched">FASTA file for non-matching database sequences</option>
|
|
|
173 <option value="--notmatched">Write non-matching query sequences to separate file</option>
|
|
|
174 <option value="--matched">Write matching query sequences to separate file</option>
|
|
|
175 </expand>
|
|
|
176
|
|
|
177 <conditional name="adv_opts">
|
|
|
178 <param name="adv_opts_selector" type="select" label="Advanced Options">
|
|
|
179 <option value="basic" selected="True">Hide Advanced Options</option>
|
|
|
180 <option value="advanced">Show Advanced Options</option>
|
|
|
181 </param>
|
|
|
182 <when value="basic" />
|
|
|
183 <when value="advanced">
|
|
|
184
|
|
|
185 <param name="top_hits_only" type="boolean" truevalue="--top_hits_only" falsevalue="" checked="False"
|
|
|
186 label="Output only hits with identity equal to the best" help="(--top_hits_only)"/>
|
|
|
187 <param name="rightjust" type="boolean" truevalue="--rightjust" falsevalue="" checked="False"
|
|
|
188 label="Reject if terminal gaps at alignment right end" help="(--rightjust)"/>
|
|
|
189 <param name="leftjust" type="boolean" truevalue="--leftjust" falsevalue="" checked="False"
|
|
|
190 label="Reject if terminal gaps at alignment left end" help="(--leftjust)"/>
|
|
|
191
|
|
|
192 <param name="query_cov" type="float" value="" optional="True" label="Reject if fraction of query seq. aligned lower than this value"
|
|
|
193 help="(--query_cov)"/>
|
|
|
194 <param name="target_cov" type="float" value="" optional="True" label="Reject if fraction of target seq. aligned lower than this value"
|
|
|
195 help="(--target_cov)"/>
|
|
|
196 <param name="maxid" type="float" value="" optional="True" label="Reject if identity higher than this value"
|
|
|
197 help="(--maxid)"/>
|
|
|
198 <param name="maxqt" type="float" value="" optional="True" label="Reject if query/target length ratio higher than this value"
|
|
|
199 help="(--maxqt)"/>
|
|
|
200 <param name="maxsizeratio" type="float" value="" optional="True" label="Reject if query/target abundance ratio higher than this value"
|
|
|
201 help="(--maxsizeratio)"/>
|
|
|
202 <param name="maxsl" type="float" value="" optional="True" label="Reject if shorter/longer length ratio higher than this value"
|
|
|
203 help="(--maxsl)"/>
|
|
|
204 <param name="mid" type="float" value="" optional="True" label="Reject if percent identity lower than this value, ignoring gaps"
|
|
|
205 help="(--mid)"/>
|
|
|
206 <param name="minqt" type="float" value="" optional="True" label="Reject if query/target length ratio lower than this value"
|
|
|
207 help="(--minqt)"/>
|
|
|
208 <param name="minsl" type="float" value="" optional="True" label="Reject if shorter/longer length ratio lower than this value"
|
|
|
209 help="(--minsl)"/>
|
|
|
210 <param name="minsizeratio" type="float" value="" optional="True" label="Reject if query/target abundance ratio lower than this value"
|
|
|
211 help="(--minsizeratio)"/>
|
|
|
212
|
|
|
213 <param name="maxqsize" type="integer" value="" optional="True" label="Reject if query abundance larger than this value"
|
|
|
214 help="(--maxqsize)"/>
|
|
|
215 <param name="mincols" type="integer" value="" optional="True" label="Reject if alignment length shorter than this value"
|
|
|
216 help="(--mincols)"/>
|
|
|
217 <param name="maxsubs" type="integer" value="" optional="True" label="Reject if more substitutions than this value"
|
|
|
218 help="(--maxsubs)"/>
|
|
|
219 <expand macro="maxrejects" />
|
|
|
220 <expand macro="maxaccepts" />
|
|
|
221 <param name="maxdiffs" type="integer" value="" optional="True" label="Reject if more substitutions or indels than this value"
|
|
|
222 help="(--maxdiffs)"/>
|
|
|
223 <param name="maxgaps" type="integer" value="" optional="True" label="Reject if more indels than this value"
|
|
|
224 help="(--maxgaps)"/>
|
|
|
225 <param name="maxhits" type="integer" value="" optional="True" label="Maximum number of hits to show"
|
|
|
226 help="(--maxhits)"/>
|
|
|
227 <param name="match" type="integer" value="2" label="Score for match"
|
|
|
228 help="(--match)"/>
|
|
|
229 <param name="idprefix" type="integer" value="" optional="True" label="Reject if first n nucleotides do not match"
|
|
|
230 help="(--idprefix)"/>
|
|
|
231 <param name="idsuffix" type="integer" value="" optional="True" label="Reject if last n nucleotides do not match"
|
|
|
232 help="(--idsuffix)"/>
|
|
|
233 <param name="mintsize" type="integer" value="" optional="True" label="Reject if target abundance lower"
|
|
|
234 help="(--mintsize)"/>
|
|
|
235 <param name="mismatch" type="integer" value="-4" label="Score for mismatch"
|
|
|
236 help="(--mismatch)"/>
|
|
|
237
|
|
|
238 <conditional name="uclust_output">
|
|
|
239 <param name="uclust_output_select" type="select" label="UCLUST-like output" help="(--uc)">
|
|
|
240 <option value="yes">Yes</option>
|
|
|
241 <option value="no" selected="True">No</option>
|
|
|
242 </param>
|
|
|
243 <when value="yes">
|
|
|
244 <param name="uc_allhits" type="boolean" truevalue="--uc_allhits" falsevalue="" checked="False"
|
|
|
245 label="Show all, not just top hits" help="(--uc_allhits)"/>
|
|
|
246 </when>
|
|
|
247 <when value="no" />
|
|
|
248 </conditional>
|
|
|
249
|
|
|
250 <param name="wordlength" type="integer" value="8" min="3" max="15" label="Length of words for database index 3-15"
|
|
|
251 help="Allowed values: 3-15 (--wordlength)"/>
|
|
|
252 <param name="output_no_hits" type="boolean" truevalue="--output_no_hits" falsevalue="" checked="False"
|
|
|
253 label="Write non-matching queries to the output file" help="(--output_no_hits)"/>
|
|
|
254
|
|
|
255 <expand macro="userfields" />
|
|
|
256 </when>
|
|
|
257 </conditional>
|
|
|
258
|
|
|
259
|
|
|
260 </inputs>
|
|
|
261 <outputs>
|
|
|
262 <data name="uc" format="fasta" label="${tool.name} on ${on_string}: UCLUST like output">
|
|
|
263 <filter>adv_opts.uclust_output.uclust_output_select == 'yes'</filter>
|
|
|
264 </data>
|
|
|
265 <data name="dbnotmatched" format="fasta" label="${tool.name} on ${on_string}: Non-matched database sequences">
|
|
|
266 <filter>'--dbnotmatchedt' in outputs</filter>
|
|
|
267 </data>
|
|
|
268 <data name="dbmatched" format="fasta" label="${tool.name} on ${on_string}: Matched database sequences">
|
|
|
269 <filter>'--dbmatched' in outputs</filter>
|
|
|
270 </data>
|
|
|
271 <data name="alnout" format="fasta" label="${tool.name} on ${on_string}: Alignment">
|
|
|
272 <filter>'--alnout' in outputs</filter>
|
|
|
273 </data>
|
|
|
274 <data name="blast6out" format="tabular" label="${tool.name} on ${on_string}: BLAST like tabular">
|
|
|
275 <filter>'--blast6out' in outputs</filter>
|
|
|
276 </data>
|
|
|
277 <data name="notmatched" format="fasta" label="${tool.name} on ${on_string}: Non-matched queries">
|
|
|
278 <filter>'--notmatched' in outputs</filter>
|
|
|
279 </data>
|
|
|
280 <data name="matched" format="fasta" label="${tool.name} on ${on_string}: Matching query sequences">
|
|
|
281 <filter>'--matched' in outputs</filter>
|
|
|
282 </data>
|
|
|
283 <data name="fastapairs" format="fasta" label="${tool.name} on ${on_string}: Query/Target sequences">
|
|
|
284 <filter>'--fastapairs' in outputs</filter>
|
|
|
285 </data>
|
|
|
286 <data name="userout" format="tabular" label="${tool.name} on ${on_string}: tabular output">
|
|
|
287 <filter>adv_opts.userfields_output.userfields_output_select == 'yes'</filter>
|
|
|
288 </data>
|
|
|
289 </outputs>
|
|
|
290 <tests>
|
|
|
291 <test>
|
|
|
292 <param name="dbfile" value="three_human_mRNA.fasta" ftype="fasta" />
|
|
|
293 <param name="queryfile" value="query.fasta" ftype="fasta" />
|
|
|
294 <param name="outputs" value="--blast6out,--dbmatched" />
|
|
|
295 <param name="dbmask" value="none" />
|
|
|
296 <param name="adv_opts_selector" value="advanced" />
|
|
|
297 <param name="top_hits_only" value="True" />
|
|
|
298 <output name="dbmatched" file="search_dbmatched_result1.fasta" ftype="fasta" />
|
|
|
299 <output name="blast6out" file="search_blast6out_result1.tabular" ftype="tabular" />
|
|
|
300 </test>
|
|
|
301 <test>
|
|
|
302 <param name="dbfile" value="three_human_mRNA.fasta" ftype="fasta" />
|
|
|
303 <param name="queryfile" value="query.fasta" ftype="fasta" />
|
|
|
304 <param name="outputs" value="--fastapairs,--matched" />
|
|
|
305 <param name="dbmask" value="none" />
|
|
|
306 <param name="adv_opts_selector" value="advanced" />
|
|
|
307 <param name="userfields_output_select" value="yes" />
|
|
|
308 <param name="userfields" value="query,target,evalue" />
|
|
|
309 <param name="top_hits_only" value="True" />
|
|
|
310 <param name="output_no_hits" value="True" />
|
|
|
311 <output name="matched" file="search_matched_result2.fasta" ftype="fasta" />
|
|
|
312 <output name="fastapairs" file="search_fastapairs_result2.fasta" ftype="fasta" />
|
|
|
313 <output name="userout" file="search_userfields_result2.tabular" ftype="tabular" />
|
|
|
314 </test>
|
|
|
315 </tests>
|
|
|
316 <help>
|
|
|
317 <![CDATA[
|
|
|
318 **What it does**
|
|
|
319
|
|
|
320 Sequence search based on vsearch.
|
|
|
321
|
|
|
322 @USERFIELDS_HELP@
|
|
|
323
|
|
|
324 Searching options
|
|
|
325 --alnout FILENAME filename for human-readable alignment output
|
|
|
326 --blast6out FILENAME filename for blast-like tab-separated output
|
|
|
327 --db FILENAME filename for FASTA formatted database for search
|
|
|
328 --dbmask mask db with "dust", "soft" or "none" method (dust)
|
|
|
329 --dbmatched FILENAME FASTA file for matching database sequences
|
|
|
330 --dbnotmatched FILENAME FASTA file for non-matching database sequences
|
|
|
331 --fastapairs FILENAME FASTA file with pairs of query and target
|
|
|
332 --fulldp full dynamic programming alignment (always on)
|
|
|
333 --gapext STRING penalties for gap extension (2I/1E)
|
|
|
334 --gapopen STRING penalties for gap opening (20I/2E)
|
|
|
335 --hardmask mask by replacing with N instead of lower case
|
|
|
336 --id REAL reject if identity lower
|
|
|
337 --iddef INT id definition, 0-4=CD-HIT,all,int,MBL,BLAST (2)
|
|
|
338 --idprefix INT reject if first n nucleotides do not match
|
|
|
339 --idsuffix INT reject if last n nucleotides do not match
|
|
|
340 --leftjust reject if terminal gaps at alignment left end
|
|
|
341 --match INT score for match (2)
|
|
|
342 --matched FILENAME FASTA file for matching query sequences
|
|
|
343 --maxaccepts INT number of hits to accept and show per strand (1)
|
|
|
344 --maxdiffs INT reject if more substitutions or indels
|
|
|
345 --maxgaps INT reject if more indels
|
|
|
346 --maxhits INT maximum number of hits to show (unlimited)
|
|
|
347 --maxid REAL reject if identity higher
|
|
|
348 --maxqsize INT reject if query abundance larger
|
|
|
349 --maxqt REAL reject if query/target length ratio higher
|
|
|
350 --maxrejects INT number of non-matching hits to consider (32)
|
|
|
351 --maxsizeratio REAL reject if query/target abundance ratio higher
|
|
|
352 --maxsl REAL reject if shorter/longer length ratio higher
|
|
|
353 --maxsubs INT reject if more substitutions
|
|
|
354 --mid REAL reject if percent identity lower, ignoring gaps
|
|
|
355 --mincols INT reject if alignment length shorter
|
|
|
356 --minqt REAL reject if query/target length ratio lower
|
|
|
357 --minsizeratio REAL reject if query/target abundance ratio lower
|
|
|
358 --minsl REAL reject if shorter/longer length ratio lower
|
|
|
359 --mintsize INT reject if target abundance lower
|
|
|
360 --mismatch INT score for mismatch (-4)
|
|
|
361 --notmatched FILENAME FASTA file for non-matching query sequences
|
|
|
362 --output_no_hits output non-matching queries to output files
|
|
|
363 --qmask mask query with "dust", "soft" or "none" method (dust)
|
|
|
364 --query_cov REAL reject if fraction of query seq. aligned lower
|
|
|
365 --rightjust reject if terminal gaps at alignment right end
|
|
|
366 --rowlen INT width of alignment lines in alnout output (64)
|
|
|
367 --self reject if labels identical
|
|
|
368 --selfid reject if sequences identical
|
|
|
369 --sizeout write abundance annotation to output
|
|
|
370 --strand search "plus" or "both" strands (plus)
|
|
|
371 --target_cov REAL reject if fraction of target seq. aligned lower
|
|
|
372 --top_hits_only output only hits with identity equal to the best
|
|
|
373 --uc FILENAME filename for UCLUST-like output
|
|
|
374 --uc_allhits show all, not just top hit with uc output
|
|
|
375 --usearch_global FILENAME filename of queries for global alignment search
|
|
|
376 --userfields STRING fields to output in userout file
|
|
|
377 --userout FILENAME filename for user-defined tab-separated output
|
|
|
378 --weak_id REAL include aligned hits with >= id; continue search
|
|
|
379 --wordlength INT length of words for database index 3-15 (8)
|
|
|
380
|
|
|
381 @EXTERNAL_DOCUMENTATION@
|
|
|
382
|
|
|
383 -------
|
|
|
384
|
|
|
385 @REFERENCES@
|
|
|
386
|
|
|
387
|
|
|
388 ]]>
|
|
|
389 </help>
|
|
|
390 <expand macro="citations" />
|
|
|
391 </tool>
|
