|
0
|
1 <tool id="directory_data_reader" name="Directory Data Finder" version="0.2">
|
|
|
2 <description>Reads a particular data type from a directory on the server</description>
|
|
|
3 <command interpreter="python">
|
|
|
4 <![CDATA[
|
|
|
5 directory_copier.py
|
|
|
6 #if $results.required=="data"
|
|
|
7 --ending .${results.extension.file_type}
|
|
|
8 --link
|
|
|
9 #if $results.start
|
|
|
10 --start $results.start
|
|
|
11 #end if
|
|
|
12 #if $results.last
|
|
|
13 --last $results.last
|
|
|
14 #end if
|
|
|
15 #if $results.extension.file_type=="fa"
|
|
|
16 --new_ending .fasta
|
|
|
17 #end if
|
|
|
18 #if $results.extension.file_type=="fq"
|
|
|
19 --new_ending .fastq
|
|
|
20 #end if
|
|
|
21 #if $results.extension.file_type=="text"
|
|
|
22 --new_ending .txt
|
|
|
23 #end if
|
|
|
24 #if $results.extension.file_type=="tsv"
|
|
|
25 --new_ending .tabular
|
|
|
26 #end if
|
|
|
27 #if $results.extension.file_type in ["fasta.gz"]
|
|
|
28 --decompress
|
|
|
29 --new_ending .fasta
|
|
|
30 #end if
|
|
|
31 #if $results.extension.file_type=="fastq"
|
|
|
32 --new_ending .$results.extension.new_galaxy.new_ending
|
|
|
33 #end if
|
|
|
34 #if $results.extension.file_type=="fastq.gz"
|
|
|
35 --decompress
|
|
|
36 --new_ending .${results.extension.new_galaxy.new_ending}
|
|
|
37 #end if
|
|
|
38 #else
|
|
|
39 --ending bam
|
|
|
40 --ending csv
|
|
|
41 --ending fa
|
|
|
42 --ending fasta
|
|
|
43 --ending fasta.gz
|
|
|
44 --ending fastq
|
|
|
45 --ending fastq.gz
|
|
|
46 --ending fasta
|
|
|
47 --ending fq
|
|
|
48 --ending sam
|
|
|
49 --ending tabular
|
|
|
50 --ending text
|
|
|
51 --ending tsv
|
|
|
52 --ending txt
|
|
|
53 --ending xls
|
|
|
54 --ending xlsx
|
|
|
55 #end if
|
|
|
56 #if $directory.startswith('/'):
|
|
|
57 --path ${directory}
|
|
|
58 #else
|
|
|
59 --path $__tool_directory__/${directory}
|
|
|
60 #end if
|
|
|
61 --list ${listing}
|
|
|
62 ]]>
|
|
|
63 </command>
|
|
|
64 <inputs>
|
|
|
65 <param name="directory" type="text" label="Directory to read data from." />
|
|
|
66 <param name="list_name" type="text" size="25" label="output name" value="input data"/>
|
|
|
67 <conditional name="results">
|
|
|
68 <param name="required" type="select" label="Download data or just directory listing" help="Select type of action required.">
|
|
|
69 <option value="data" selected="true">Data and listing of selected type</option>
|
|
|
70 <option value="listing">Get listing of selected file types </option>
|
|
|
71 </param>
|
|
|
72 <when value="data">
|
|
|
73 <param name="start" type="text" value="" label="String which must be at the start of each file name" />
|
|
|
74 <param name="last" type="text" value="" label="String which must be at the end of the file name (excluding the file type)" />
|
|
|
75 <conditional name="extension">
|
|
|
76 <param name="file_type" type="select" label="File Type" help="File Type.">
|
|
|
77 <option value="bam">*.bam files</option>
|
|
|
78 <option value="csv">*.csv files</option>
|
|
|
79 <option value="fa">*.fa files Files saved as *.fasta for galaxy</option>
|
|
|
80 <option value="fasta">*.fasta files</option>
|
|
|
81 <option value="fasta.gz">*.fasta.gz files</option>
|
|
|
82 <option value="fastq">*.fastq files</option>
|
|
|
83 <option value="fastq.gz">*.fastq.gz files</option>
|
|
|
84 <option value="fastq">*.fastq files</option>
|
|
|
85 <option value="fastq.gz">*.fastq.gz files</option>
|
|
|
86 <option value="fq">*.fq files Files saved as *.fastq for galaxy</option>
|
|
|
87 <option value="sam">*.sam files</option>
|
|
|
88 <option value="tabular">*.tabular Files</option>
|
|
|
89 <option value="text">*.text Files saved as *.txt for galaxy</option>
|
|
|
90 <option value="tsv">*.tsv files saved as *.tabular for galaxy</option>
|
|
|
91 <option value="txt">*.txt Files</option>
|
|
|
92 <option value="xls">*.xls files</option>
|
|
|
93 <option value="xlsx">*.xlsx files</option>
|
|
|
94 </param>
|
|
|
95 <when value="bam" />
|
|
|
96 <when value="csv" />
|
|
|
97 <when value="fa" />
|
|
|
98 <when value="fasta" />
|
|
|
99 <when value="fasta.gz" />
|
|
|
100 <when value="fastq" >
|
|
|
101 <conditional name="new_galaxy">
|
|
|
102 <param name="new_ending" type="select" label="Ending to be used for Galaxy" help="Will determine which down stream tools can be used.">
|
|
|
103 <option value="fastq" selected="true">Keep data as general fastq format</option>
|
|
|
104 <option value="fastqsanger">Tag data as fastq sanger in galaxy</option>
|
|
|
105 <option value="fastqsolexa">Tag data as fastq solexa in galaxy</option>
|
|
|
106 <option value="fastqillumina">Tag data as fastq illumina in galaxy</option>
|
|
|
107 </param>
|
|
|
108 <when value="fastq" />
|
|
|
109 <when value="fastqsanger" />
|
|
|
110 <when value="fastqsolexa" />
|
|
|
111 <when value="fastqillumina" />
|
|
|
112 </conditional>
|
|
|
113 </when>
|
|
|
114 <when value="fastq.gz" >
|
|
|
115 <conditional name="new_galaxy">
|
|
|
116 <param name="new_ending" type="select" label="Ending to be used for Galaxy" help="Will determine which down stream tools can be used.">
|
|
|
117 <option value="fastq" selected="true" >Keep data as general fastq format</option>
|
|
|
118 <option value="fastqsanger">Tag data as fastq sanger in galaxy</option>
|
|
|
119 <option value="fastqsolexa">Tag data as fastq solexa in galaxy</option>
|
|
|
120 <option value="fastqillumina">Tag data as fastq illumina in galaxy</option>
|
|
|
121 </param>
|
|
|
122 <when value="fastq" />
|
|
|
123 <when value="fastqsanger" />
|
|
|
124 <when value="fastqsolexa" />
|
|
|
125 <when value="fastqillumina" />
|
|
|
126 </conditional>
|
|
|
127 </when>
|
|
|
128 <when value="fq" />
|
|
|
129 <when value="sam" />
|
|
|
130 <when value="tabular" />
|
|
|
131 <when value="text" />
|
|
|
132 <when value="tsv" />
|
|
|
133 <when value="txt" />
|
|
|
134 <when value="xls" />
|
|
|
135 <when value="xlsx" />
|
|
|
136 </conditional>
|
|
|
137 </when>
|
|
|
138 <when value="listing">
|
|
|
139 </when>
|
|
|
140 </conditional>
|
|
|
141 </inputs>
|
|
|
142 <outputs>
|
|
|
143 <data format="txt" name="listing" label="List of files in $list_name">
|
|
|
144 </data>
|
|
|
145 <!-- Ideally galaxy can get the type based on the file extensions. If so just add the type here -->
|
|
|
146 <collection type="list" label="$list_name" name="data_collection">
|
|
|
147 <filter>(results['required'] == 'data')</filter>
|
|
|
148 <discover_datasets pattern="__designation_and_ext__" directory="output" visible="true" />
|
|
|
149 </collection>
|
|
|
150 </outputs>
|
|
|
151 <tests>
|
|
|
152 <test>
|
|
|
153 <param name="directory" value="test-data" />
|
|
|
154 <param name="list_name" value="csv_files" />
|
|
|
155 <param name="results|extension|file_type" value="csv"/>
|
|
|
156 <output name="listing">
|
|
|
157 <assert_contents>
|
|
|
158 <has_line line="sample1.csv" />
|
|
|
159 </assert_contents>
|
|
|
160 </output>
|
|
|
161 <output_collection name="data_collection" type="list">
|
|
|
162 <element name="sample1" ftype="csv" file="sample1.csv" />
|
|
|
163 </output_collection>
|
|
|
164 </test>
|
|
|
165 <test>
|
|
|
166 <param name="directory" value="test-data" />
|
|
|
167 <param name="results|extension|file_type" value="fa"/>
|
|
|
168 <output name="listing">
|
|
|
169 <assert_contents>
|
|
|
170 <has_line line="sample1.fasta" />
|
|
|
171 <has_line line="other.fasta" />
|
|
|
172 </assert_contents>
|
|
|
173 </output>
|
|
|
174 <output_collection name="data_collection" type="list">
|
|
|
175 <element name="sample1" ftype="fasta" file="sample1.fasta" />
|
|
|
176 <element name="other" ftype="fasta" file="sample1.fasta" />
|
|
|
177 </output_collection>
|
|
|
178 </test>
|
|
|
179 <test>
|
|
|
180 <param name="directory" value="test-data" />
|
|
|
181 <param name="results|extension|file_type" value="fasta"/>
|
|
|
182 <param name="results|start" value="sam" />
|
|
|
183 <output name="listing">
|
|
|
184 <assert_contents>
|
|
|
185 <has_line line="sample1.fasta" />
|
|
|
186 <not_has_text text="other.fasta" />
|
|
|
187 </assert_contents>
|
|
|
188 </output>
|
|
|
189 <output_collection name="data_collection" type="list">
|
|
|
190 <element name="sample1" ftype="fasta" file="sample1.fasta" />
|
|
|
191 </output_collection>
|
|
|
192 </test>
|
|
|
193 <test>
|
|
|
194 <param name="directory" value="test-data" />
|
|
|
195 <param name="results|extension|file_type" value="fasta.gz"/>
|
|
|
196 <param name="results|last" value="le1" />
|
|
|
197 <output name="listing">
|
|
|
198 <assert_contents>
|
|
|
199 <has_line line="sample1.fasta" />
|
|
|
200 <not_has_text text="other.fasta" />
|
|
|
201 </assert_contents>
|
|
|
202 </output>
|
|
|
203 <output_collection name="data_collection" type="list">
|
|
|
204 <element name="sample1" ftype="fasta" file="sample1.fasta" />
|
|
|
205 </output_collection>
|
|
|
206 </test>
|
|
|
207 <test>
|
|
|
208 <param name="directory" value="test-data" />
|
|
|
209 <param name="results|extension|file_type" value="fq"/>
|
|
|
210 <output name="listing">
|
|
|
211 <assert_contents>
|
|
|
212 <has_line line="sample1.fastq" />
|
|
|
213 </assert_contents>
|
|
|
214 </output>
|
|
|
215 <output_collection name="data_collection" type="list">
|
|
|
216 <element name="sample1" ftype="fastq" file="sample1.fastq" />
|
|
|
217 </output_collection>
|
|
|
218 </test>
|
|
|
219 <test>
|
|
|
220 <param name="directory" value="test-data" />
|
|
|
221 <param name="results|extension|file_type" value="fastq"/>
|
|
|
222 <output name="listing">
|
|
|
223 <assert_contents>
|
|
|
224 <has_line line="sample1.fastq" />
|
|
|
225 </assert_contents>
|
|
|
226 </output>
|
|
|
227 <output_collection name="data_collection" type="list">
|
|
|
228 <element name="sample1" ftype="fastq" file="sample1.fastq" />
|
|
|
229 </output_collection>
|
|
|
230 </test>
|
|
|
231 <test>
|
|
|
232 <param name="directory" value="test-data" />
|
|
|
233 <param name="results|extension|file_type" value="fastq"/>
|
|
|
234 <param name="results|extension|new_galaxy|new_ending" value="fastqsanger"/>
|
|
|
235 <output name="listing">
|
|
|
236 <assert_contents>
|
|
|
237 <has_line line="sample1.fastqsanger" />
|
|
|
238 </assert_contents>
|
|
|
239 </output>
|
|
|
240 <output_collection name="data_collection" type="list">
|
|
|
241 <element name="sample1" ftype="fastqsanger" file="sample1.fastq" />
|
|
|
242 </output_collection>
|
|
|
243 </test>
|
|
|
244 <test>
|
|
|
245 <param name="directory" value="test-data" />
|
|
|
246 <param name="results|extension|file_type" value="fastq.gz"/>
|
|
|
247 <output name="data_collection">
|
|
|
248 <assert_contents>
|
|
|
249 <has_line line="sample1.fastq" />
|
|
|
250 </assert_contents>
|
|
|
251 </output>
|
|
|
252 <output_collection name="data_collection" type="list">
|
|
|
253 <element name="sample1" ftype="fastq" file="sample1.fastq" />
|
|
|
254 </output_collection>
|
|
|
255 </test>
|
|
|
256 <test>
|
|
|
257 <param name="directory" value="test-data" />
|
|
|
258 <param name="results|extension|file_type" value="fastq.gz"/>
|
|
|
259 <param name="results|extension|new_galaxy|new_ending" value="fastqsanger"/>
|
|
|
260 <output name="listing_fastq_gz">
|
|
|
261 <assert_contents>
|
|
|
262 <has_line line="sample1.fastqsanger" />
|
|
|
263 </assert_contents>
|
|
|
264 </output>
|
|
|
265 <output_collection name="data_collection" type="list">
|
|
|
266 <element name="sample1" ftype="fastqsanger" file="sample1.fastq" />
|
|
|
267 </output_collection>
|
|
|
268 </test>
|
|
|
269 <test>
|
|
|
270 <param name="directory" value="test-data" />
|
|
|
271 <param name="results|extension|file_type" value="sam"/>
|
|
|
272 <output name="listing">
|
|
|
273 <assert_contents>
|
|
|
274 <has_line line="sample1.sam" />
|
|
|
275 </assert_contents>
|
|
|
276 </output>
|
|
|
277 <output_collection name="data_collection" type="list">
|
|
|
278 <element name="sample1" ftype="sam" file="sample1.sam" />
|
|
|
279 </output_collection>
|
|
|
280 </test>
|
|
|
281 <test>
|
|
|
282 <param name="directory" value="test-data" />
|
|
|
283 <param name="results|extension|file_type" value="tabular"/>
|
|
|
284 <output name="listing">
|
|
|
285 <assert_contents>
|
|
|
286 <has_line line="sample1.tabular" />
|
|
|
287 </assert_contents>
|
|
|
288 </output>
|
|
|
289 <output_collection name="data_collection" type="list">
|
|
|
290 <element name="sample1" ftype="tabular" file="sample1.tabular" />
|
|
|
291 </output_collection>
|
|
|
292 </test>
|
|
|
293 <test>
|
|
|
294 <param name="directory" value="test-data" />
|
|
|
295 <param name="results|extension|file_type" value="text"/>
|
|
|
296 <output name="listing_text">
|
|
|
297 <assert_contents>
|
|
|
298 <has_line line="sample1.txt" />
|
|
|
299 </assert_contents>
|
|
|
300 </output>
|
|
|
301 <output_collection name="data_collection" type="list">
|
|
|
302 <element name="sample1" ftype="txt" file="sample1.text" />
|
|
|
303 </output_collection>
|
|
|
304 </test>
|
|
|
305 <test>
|
|
|
306 <param name="directory" value="test-data" />
|
|
|
307 <param name="results|extension|file_type" value="tsv"/>
|
|
|
308 <output name="data_collection">
|
|
|
309 <assert_contents>
|
|
|
310 <has_line line="sample1.tabular" />
|
|
|
311 </assert_contents>
|
|
|
312 </output>
|
|
|
313 <output_collection name="data_collection" type="list">
|
|
|
314 <element name="sample1" ftype="tabular" file="sample1.tsv" />
|
|
|
315 </output_collection>
|
|
|
316 </test>
|
|
|
317 <test>
|
|
|
318 <param name="directory" value="test-data" />
|
|
|
319 <param name="results|extension|file_type" value="txt" />
|
|
|
320 <output name="listing">
|
|
|
321 <assert_contents>
|
|
|
322 <has_line line="sample1.txt" />
|
|
|
323 </assert_contents>
|
|
|
324 </output>
|
|
|
325 <output_collection name="data_collection" type="list">
|
|
|
326 <element name="sample1" ftype="txt" file="sample1.txt" />
|
|
|
327 <element name="sample2" ftype="txt" file="sample2.txt" />
|
|
|
328 </output_collection>
|
|
|
329 </test>
|
|
|
330 <test>
|
|
|
331 <param name="directory" value="test-data" />
|
|
|
332 <param name="results|extension|file_type" value="xls"/>
|
|
|
333 <output name="listing">
|
|
|
334 <assert_contents>
|
|
|
335 <has_line line="sample1.xls" />
|
|
|
336 </assert_contents>
|
|
|
337 </output>
|
|
|
338 <output_collection name="data_collection" type="list">
|
|
|
339 <element name="sample1" ftype="xls" file="sample1.xls" />
|
|
|
340 </output_collection>
|
|
|
341 </test>
|
|
|
342 <test>
|
|
|
343 <param name="directory" value="test-data" />
|
|
|
344 <param name="results|extension|file_type" value="xlsx"/>
|
|
|
345 <output name="listing">
|
|
|
346 <assert_contents>
|
|
|
347 <has_line line="sample1.xlsx" />
|
|
|
348 </assert_contents>
|
|
|
349 </output>
|
|
|
350 <output_collection name="data_collection" type="list">
|
|
|
351 <element name="sample1" ftype="xlsx" file="sample1.xlsx" />
|
|
|
352 </output_collection>
|
|
|
353 </test>
|
|
|
354 <test>
|
|
|
355 <param name="directory" value="test-data" />
|
|
|
356 <param name="results|required" value="listing"/>
|
|
|
357 <output name="listing_all">
|
|
|
358 <assert_contents>
|
|
|
359 <has_line line="sample1.csv" />
|
|
|
360 <has_line line="sample1.fasta" />
|
|
|
361 <has_line line="sample1.fasta.gz" />
|
|
|
362 <has_line line="sample1.fastq" />
|
|
|
363 <has_line line="sample1.fastq.gz" />
|
|
|
364 <has_line line="sample1.sam" />
|
|
|
365 <has_line line="sample1.tabular" />
|
|
|
366 <has_line line="sample1.text" />
|
|
|
367 <has_line line="sample1.tsv" />
|
|
|
368 <has_line line="sample1.txt" />
|
|
|
369 <has_line line="sample1.xls" />
|
|
|
370 <has_line line="sample1.xlsx" />
|
|
|
371 </assert_contents>
|
|
|
372 </output>
|
|
|
373 </test>
|
|
|
374
|
|
|
375 </tests>
|
|
|
376
|
|
|
377 <help>
|
|
|
378 <![CDATA[
|
|
|
379 This tool will lookup files on the Galaxy server machine, including mounted directories.
|
|
|
380
|
|
|
381 Only directories that are included in the white list and not in the black list are allowed.
|
|
|
382 If the directory you require does not pass the white list or blacklist test please contact the local galaxy admin.
|
|
|
383 (Admins see README)
|
|
|
384
|
|
|
385 This tool only supports a limited set of types and file extenstions.
|
|
|
386 No other files are ever returned either by data or listing.
|
|
|
387 These endings are case senitive.
|
|
|
388
|
|
|
389 ====
|
|
|
390
|
|
|
391 The data options will look for all files that have a particular ending in the selected directory.
|
|
|
392
|
|
|
393 The tool will return two things.
|
|
|
394
|
|
|
395 1. A Dataset collection of all the detected files.
|
|
|
396
|
|
|
397 2. A file with the names of all the detected files. These will be sorted in the same order as galaxy builds the dataset collection.
|
|
|
398
|
|
|
399 The files can be filtered by setting a specific start strinf for the file name.
|
|
|
400 Only files that start with this string (case senstive) will be included.
|
|
|
401
|
|
|
402 Files can also be filter for the last part before the file extsentions.
|
|
|
403
|
|
|
404 Assuming the directory has:
|
|
|
405 C01_R1_001.fasta C01_R2_001.fatsa C02_R1_001.fasta C02_R2_001.fatsa
|
|
|
406
|
|
|
407 Setting start C01 will return just the C01 files: C01_R1_001.fasta C01_R2_001.fatsa
|
|
|
408
|
|
|
409 Setting last R1_001 will return the read1 files: C01_R1_001.fasta C02_R1_001.fasta
|
|
|
410
|
|
|
411 As Galaxy detects the file type based on the extension this tool will change the exstension for supported alternative file ends.
|
|
|
412 This includes (manually) setting the exstension to fastqsanger, fastqsolexa, fastqillumina for tools that specify one of these.
|
|
|
413
|
|
|
414 This tool will unzip gz files.
|
|
|
415
|
|
|
416 ====
|
|
|
417
|
|
|
418 The listing option will return a txt file with all the files found with any of the supported endings. Other files in that directory are not included.
|
|
|
419
|
|
|
420 The file exstensions are left as in the directory.
|
|
|
421
|
|
|
422 File start and last filters are not supported in this mode.
|
|
|
423 ]]>
|
|
|
424 </help>
|
|
|
425 <citations>
|
|
|
426 </citations>
|
|
|
427
|
|
|
428 </tool>
|
