Mercurial > repos > nml > csvtk_summary

comparison summary.xml @ 0:ceb70f0dd898 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/shenwei356/csvtk commit 3a97e1b79bf0c6cdd37d5c8fb497b85531a563ab"
author nml
date Tue, 19 May 2020 17:23:57 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:ceb70f0dd898
1 <tool id="csvtk_summary" name="csvtk-summary" version="@VERSION@+@GALAXY_VERSION@">
2 <description> statistics of selected fields</description>
3 <macros>
4 <import>macros.xml</import>
5 </macros>
6 <expand macro="requirements" />
7 <expand macro="version_cmd" />
8 <command detect_errors="exit_code"><![CDATA[
9
10 ## Set Up Input ##
11 ##################
12
13 #set input_list = list()
14 #for $repeat in $field
15 #silent $input_list.append(str($repeat.column_text.in_text) + ":" + str($repeat.analysis_type))
16 #end for
17
18 #set input_total = ",".join($input_list)
19
20 ###################
21 ## Start Command ##
22 ###################
23
24 csvtk summary --num-cpus "\${GALAXY_SLOTS:-1}"
25
26 ## Add additional flags as specified ##
27 #######################################
28 $global_param.illegal_rows
29 $global_param.empty_rows
30 $global_param.header
31 $global_param.lazy_quotes
32
33 ## Set Tabular input/output flag if input is tabular ##
34 #######################################################
35 #if $in_1.is_of_type("tabular"):
36 -t -T
37 #end if
38
39 ## Set Input ##
40 ###############
41 '$in_1'
42
43 ## Specify fields ##
44 ####################
45 -f '$input_total'
46
47 ## other ##
48 ###########
49
50 #if $group_field.select_group != "none"
51 -g '$group_field.in_text'
52 #end if
53
54 -s '$extra.separator'
55 -S '$extra.rand_int'
56 -n '$decimal_width'
57 $ignore_non_digits
58
59
60 ## To output ##
61 ###############
62 > summary
63
64 ]]></command>
65 <inputs>
66 <expand macro="singular_input" />
67 <repeat name="field" title="Select Column and Operator" min="1">
68 <expand macro="singular_fields_input" />
69 <param name="analysis_type" type="select" label="Analysis Type" help="Select analysis type to do on the chosen field">
70 <option value="collapse">Collapse</option>
71 <option value="count">Count</option>
72 <option value="countn">Count of Digits (countn)</option>
73 <option value="countunique">Count Unique</option>
74 <option value="first">First Value</option>
75 <option value="last">Last Value</option>
76 <option value="max">Maximum</option>
77 <option value="mean">Mean</option>
78 <option value="median">Median</option>
79 <option value="min">Minimum</option>
80 <option value="prod">Product of the Elements</option>
81 <option value="q1">q1</option>
82 <option value="q2">q2</option>
83 <option value="q3">q3</option>
84 <option value="rand">Random Value</option>
85 <option value="entropy">Shannon Entropy</option>
86 <option value="stdev">Standard Deviation</option>
87 <option value="sum">Sum</option>
88 <option value="uniq">Unique</option>
89 <option value="variance">Variance</option>
90 </param>
91 </repeat>
92 <expand macro="groups_input" />
93 <param name="decimal_width" type="integer" value="2" argument="-n"
94 label="Number of Decimals"
95 help="Limit float to N decimal places"
96 />
97 <param name="ignore_non_digits" type="boolean" checked="false" argument="-i"
98 truevalue="-i"
99 falsevalue=""
100 label="Ignore non-digits"
101 help="Ignore non-digit values in columns. Ex. NA or N/A"
102 />
103 <section name="extra" title="Specific Optional Analysis Modifiers" expanded="false">
104 <param name="separator" type="text" value="; " argument="-s"
105 label="Collapse Separator String"
106 help="Input string of characters that will separate collapsed columns. The ' character is not allowed">
107 <expand macro="text_sanitizer" />
108 </param>
109 <param name="rand_int" type="integer" value="11" argument="-S"
110 label="Random Value Seed"
111 help="specify an integer"
112 />
113 </section>
114 <expand macro="global_parameters" />
115 </inputs>
116 <outputs>
117 <data format_source="in_1" name="summary" from_work_dir="summary" label="${in_1.name} summary of analyses" />
118 </outputs>
119 <tests>
120 <test>
121 <param name="in_1" value="plot.csv" />
122 <repeat name="field">
123 <conditional name="column_text">
124 <param name="select" value="string" />
125 <param name="in_text" value="2" />
126 </conditional>
127 <param name="analysis_type" value="collapse" />
128 </repeat>
129 <repeat name="field">
130 <conditional name="column_text">
131 <param name="select" value="string" />
132 <param name="in_text" value="3" />
133 </conditional>
134 <param name="analysis_type" value="count" />
135 </repeat>
136 <repeat name="field">
137 <conditional name="column_text">
138 <param name="select" value="string" />
139 <param name="in_text" value="2" />
140 </conditional>
141 <param name="analysis_type" value="countn" />
142 </repeat>
143 <repeat name="field">
144 <conditional name="column_text">
145 <param name="select" value="string" />
146 <param name="in_text" value="3" />
147 </conditional>
148 <param name="analysis_type" value="countunique" />
149 </repeat>
150 <repeat name="field">
151 <conditional name="column_text">
152 <param name="select" value="string" />
153 <param name="in_text" value="2" />
154 </conditional>
155 <param name="analysis_type" value="entropy" />
156 </repeat>
157 <repeat name="field">
158 <conditional name="column_text">
159 <param name="select" value="string" />
160 <param name="in_text" value="3" />
161 </conditional>
162 <param name="analysis_type" value="first" />
163 </repeat>
164 <repeat name="field">
165 <conditional name="column_text">
166 <param name="select" value="string" />
167 <param name="in_text" value="2" />
168 </conditional>
169 <param name="analysis_type" value="last" />
170 </repeat>
171 <repeat name="field">
172 <conditional name="column_text">
173 <param name="select" value="string" />
174 <param name="in_text" value="3" />
175 </conditional>
176 <param name="analysis_type" value="max" />
177 </repeat>
178 <repeat name="field">
179 <conditional name="column_text">
180 <param name="select" value="string" />
181 <param name="in_text" value="2" />
182 </conditional>
183 <param name="analysis_type" value="mean" />
184 </repeat>
185 <repeat name="field">
186 <conditional name="column_text">
187 <param name="select" value="string" />
188 <param name="in_text" value="3" />
189 </conditional>
190 <param name="analysis_type" value="median" />
191 </repeat>
192 <repeat name="field">
193 <conditional name="column_text">
194 <param name="select" value="string" />
195 <param name="in_text" value="2" />
196 </conditional>
197 <param name="analysis_type" value="min" />
198 </repeat>
199 <repeat name="field">
200 <conditional name="column_text">
201 <param name="select" value="string" />
202 <param name="in_text" value="3" />
203 </conditional>
204 <param name="analysis_type" value="prod" />
205 </repeat>
206 <repeat name="field">
207 <conditional name="column_text">
208 <param name="select" value="string" />
209 <param name="in_text" value="2" />
210 </conditional>
211 <param name="analysis_type" value="q1" />
212 </repeat>
213 <repeat name="field">
214 <conditional name="column_text">
215 <param name="select" value="string" />
216 <param name="in_text" value="3" />
217 </conditional>
218 <param name="analysis_type" value="q2" />
219 </repeat>
220 <repeat name="field">
221 <conditional name="column_text">
222 <param name="select" value="string" />
223 <param name="in_text" value="2" />
224 </conditional>
225 <param name="analysis_type" value="q3" />
226 </repeat>
227 <repeat name="field">
228 <conditional name="column_text">
229 <param name="select" value="string" />
230 <param name="in_text" value="3" />
231 </conditional>
232 <param name="analysis_type" value="rand" />
233 </repeat>
234 <repeat name="field">
235 <conditional name="column_text">
236 <param name="select" value="string" />
237 <param name="in_text" value="2" />
238 </conditional>
239 <param name="analysis_type" value="stdev" />
240 </repeat>
241 <repeat name="field">
242 <conditional name="column_text">
243 <param name="select" value="string" />
244 <param name="in_text" value="3" />
245 </conditional>
246 <param name="analysis_type" value="sum" />
247 </repeat>
248 <repeat name="field">
249 <conditional name="column_text">
250 <param name="select" value="string" />
251 <param name="in_text" value="2" />
252 </conditional>
253 <param name="analysis_type" value="uniq" />
254 </repeat>
255 <repeat name="field">
256 <conditional name="column_text">
257 <param name="select" value="string" />
258 <param name="in_text" value="3" />
259 </conditional>
260 <param name="analysis_type" value="variance" />
261 </repeat>
262 <conditional name="group_field">
263 <param name="select_group" value="string" />
264 <param name="in_text" value="1" />
265 </conditional>
266 <output name="summary" >
267 <assert_contents>
268 <has_text text="collapse" />
269 <has_text text="count" />
270 <has_text text="countn" />
271 <has_text text="countunique" />
272 <has_text text="entropy" />
273 <has_text text="first" />
274 <has_text text="last" />
275 <has_text text="max" />
276 <has_text text="mean" />
277 <has_text text="median" />
278 <has_text text="min" />
279 <has_text text="prod" />
280 <has_text text="q1" />
281 <has_text text="q2" />
282 <has_text text="q3" />
283 <has_text text="rand" />
284 <has_text text="stdev" />
285 <has_text text="sum" />
286 <has_text text="uniq" />
287 <has_text text="variance" />
288 </assert_contents>
289 </output>
290 </test>
291 </tests>
292 <help><![CDATA[
293
294 Csvtk - Summary Help
295 --------------------
296
297 Info
298 ####
299
300 Csvtk Summary works to allow the use of a variety of analysis tools on the selected columns(s) and display one output at the end
301
302 .. class:: warningmark
303
304 Single quotes are not allowed in text inputs!
305
306 @HELP_INPUT_DATA@
307
308
309 Usage
310 #####
311
312 To run csvtk-summary, all you need is a valid (as defined above) CSV or TSV file with any column(s) that you want to
313 run one of the analyses on.
314
315 Analyses include:
316
317 - Collapse
318
319 - Count
320
321 - Count Numbers (countn)
322
323 - Count Unique
324
325 - First Value Selection
326
327 - Last Value Selection
328
329 - Maximum
330
331 - Mean
332
333 - Median
334
335 - Minimum
336
337 - q1
338
339 - q2
340
341 - q3
342
343 - Random Value Selection
344
345 - Shannon Entropy
346
347 - Sum
348
349 - Unique Values
350
351 - Variance
352
353 More information on these can be found on the `csvtk website. <https://bioinf.shenwei.me/csvtk/usage/#summary>`_
354
355 **Example Summary Input**
356
357 Input table:
358
359 +-------+--------+
360 | Group | Length |
361 +=======+========+
362 | A | 1500 |
363 +-------+--------+
364 | B | 1000 |
365 +-------+--------+
366 | B | 1500 |
367 +-------+--------+
368 | B | 2000 |
369 +-------+--------+
370
371 Suppose you wanted to group the values based on column 1 of the input table and then find out the mean lenght and maximum length for each group.
372 You would input this into csvtk-summary by creating 2 input repeats where the first one selects "column 2" and an analysis of "mean" and the
373 second one selects "column 2" with an analysis of "maximum".
374
375 Running this would generate the following output:
376
377 +-------+-------------+------------+
378 | Group | Length:mean | Length:max |
379 +=======+=============+============+
380 | A | 1500 | 1500 |
381 +-------+-------------+------------+
382 | B | 1500 | 2000 |
383 +-------+-------------+------------+
384
385 --------
386
387
388 @HELP_COLUMNS@
389
390
391 @HELP_END_STATEMENT@
392
393
394 ]]></help>
395 <expand macro="citations" />
396 </tool>