|
71
|
1 <tool id="plant_tribes_gene_family_classifier" name="Classify gene sequences" version="0.4">
|
|
28
|
2 <description>into precomputed orthologous gene family clusters</description>
|
|
0
|
3 <requirements>
|
|
71
|
4 <requirement type="package" version="0.4">plant_tribes_gene_family_classifier</requirement>
|
|
0
|
5 </requirements>
|
|
|
6 <stdio>
|
|
|
7 <!-- Anything other than zero is an error -->
|
|
|
8 <exit_code range="1:" />
|
|
29
|
9 <exit_code range=":-1" />
|
|
|
10 <!-- In case the return code has not been set propery check stderr too -->
|
|
0
|
11 <regex match="Error:" />
|
|
|
12 <regex match="Exception:" />
|
|
|
13 </stdio>
|
|
|
14 <command>
|
|
|
15 <![CDATA[
|
|
39
|
16 #import os
|
|
35
|
17
|
|
21
|
18 #if str($options_type.options_type_selector) == 'advanced':
|
|
51
|
19 #set specify_super_orthogroups_cond = $options_type.specify_super_orthogroups_cond
|
|
|
20 #set specify_super_orthogroups = $specify_super_orthogroups_cond.specify_super_orthogroups
|
|
17
|
21 #set create_orthogroup_cond = $options_type.create_orthogroup_cond
|
|
|
22 #set create_orthogroup = $create_orthogroup_cond.create_orthogroup
|
|
44
|
23 #set specify_single_copy_cond = $options_type.specify_single_copy_cond
|
|
46
|
24 #set specify_single_copy = $specify_single_copy_cond.specify_single_copy
|
|
51
|
25 #if str($specify_super_orthogroups) == 'yes':
|
|
|
26 #set specify_super_orthos = True
|
|
|
27 #set super_orthogroups = $specify_super_orthogroups_cond.super_orthogroups
|
|
|
28 #else:
|
|
|
29 #set specify_super_orthos = False
|
|
|
30 #end if
|
|
21
|
31 #if str($create_orthogroup) == 'yes':
|
|
44
|
32 #set create_ortho_sequences = True
|
|
21
|
33 #set create_corresponding_coding_sequences_cond = $create_orthogroup_cond.create_corresponding_coding_sequences_cond
|
|
|
34 #if str($create_corresponding_coding_sequences_cond.create_corresponding_coding_sequences) == 'yes':
|
|
|
35 #set create_corresponding_coding_sequences = True
|
|
|
36 #else:
|
|
|
37 #set create_corresponding_coding_sequences = False
|
|
44
|
38 #end if
|
|
21
|
39 #else:
|
|
|
40 #set create_ortho_sequences = False
|
|
23
|
41 #set create_corresponding_coding_sequences = False
|
|
21
|
42 #end if
|
|
52
|
43 #if str($specify_single_copy) == 'yes':
|
|
|
44 #set single_copy_orthogroup = True
|
|
|
45 #set single_copy_cond = $specify_single_copy_cond.single_copy_cond
|
|
|
46 #set single_copy = $single_copy_cond.single_copy
|
|
|
47 #else:
|
|
|
48 #set single_copy_orthogroup = False
|
|
|
49 #end if
|
|
13
|
50 #else:
|
|
47
|
51 #set single_copy_orthogroup = False
|
|
13
|
52 #set create_ortho_sequences = False
|
|
23
|
53 #set create_corresponding_coding_sequences = False
|
|
4
|
54 #end if
|
|
20
|
55
|
|
98
|
56 python $__tool_directory__/gene_family_classifier.py
|
|
100
|
57 --input '$input'
|
|
67
|
58 --scaffold '$scaffold.fields.path'
|
|
0
|
59 --method $method
|
|
13
|
60 --classifier $save_hmmscan_log_cond.classifier
|
|
69
|
61 --config_dir '$scaffold.fields.path'
|
|
30
|
62 --num_threads \${GALAXY_SLOTS:-4}
|
|
20
|
63
|
|
0
|
64 #if str($options_type.options_type_selector) == 'advanced':
|
|
51
|
65 #if specify_super_orthos:
|
|
|
66 --super_orthogroups $super_orthogroups
|
|
|
67 #end if
|
|
47
|
68 #if $single_copy_orthogroup:
|
|
68
|
69 #if str($single_copy) == 'custom':
|
|
70
|
70 #set single_copy_custom_cond = $single_copy_cond.single_copy_custom_cond
|
|
|
71 #set single_copy_custom = $single_copy_custom_cond.single_copy_custom
|
|
|
72 #if str($single_copy_custom) == 'no':
|
|
72
|
73 --single_copy_custom 'default'
|
|
70
|
74 #else:
|
|
|
75 --single_copy_custom '$single_copy_custom_cond.single_copy_custom_config'
|
|
|
76 #end if
|
|
68
|
77 #else:
|
|
|
78 --single_copy_taxa $single_copy_cond.single_copy_taxa
|
|
|
79 --taxa_present $single_copy_cond.taxa_present
|
|
|
80 #end if
|
|
0
|
81 #end if
|
|
89
|
82 #if $create_ortho_sequences:
|
|
99
|
83 --orthogroup_fasta "true"
|
|
18
|
84 #if $create_corresponding_coding_sequences:
|
|
35
|
85 --coding_sequences '$create_corresponding_coding_sequences_cond.coding_sequences'
|
|
17
|
86 #end if
|
|
0
|
87 #end if
|
|
|
88 #end if
|
|
20
|
89
|
|
97
|
90 #if str($save_hmmscan_log_cond.classifier) == 'hmmscan' or str($save_hmmscan_log_cond.classifier) == 'both' and str($save_hmmscan_log_cond.save_hmmscan_log) == 'yes':
|
|
|
91 --save_hmmscan_log
|
|
|
92 --hmmscan_log '$hmmscan_log'
|
|
12
|
93 #end if
|
|
58
|
94 #if $create_ortho_sequences:
|
|
62
|
95 #if $create_corresponding_coding_sequences:
|
|
97
|
96 --output_ptorthocs '$output_ptorthocs'
|
|
102
|
97 --output_ptorthocs_dir '$output_ptorthocs.files_path'
|
|
62
|
98 #else:
|
|
97
|
99 --output_ptortho '$output_ptortho'
|
|
102
|
100 --output_ptortho_dir '$output_ptortho.files_path'
|
|
18
|
101 #end if
|
|
4
|
102 #end if
|
|
47
|
103 #if $single_copy_orthogroup:
|
|
58
|
104 #if $create_ortho_sequences:
|
|
97
|
105 --output_ptsco '$output_ptsco'
|
|
102
|
106 --output_ptsco_dir '$output_ptsco.files_path'
|
|
48
|
107 #end if
|
|
44
|
108 #end if
|
|
0
|
109 ]]>
|
|
|
110 </command>
|
|
|
111 <inputs>
|
|
|
112 <param name="input" format="fasta" type="data" label="Amino acids (proteins) sequences fasta file"/>
|
|
|
113 <param name="scaffold" type="select" label="Orthogroups or gene families proteins scaffold">
|
|
|
114 <options from_data_table="plant_tribes_scaffolds" />
|
|
39
|
115 <validator type="no_options" message="No PlantTribes scaffolds are available. Use the PlantTribes Scaffolds Download Data Manager tool in Galaxy to install and populate the PlantTribes scaffolds data table."/>
|
|
0
|
116 </param>
|
|
|
117 <param name="method" type="select" label="Protein clustering method">
|
|
|
118 <option value="gfam" selected="true">GFam</option>
|
|
|
119 <option value="orthofinder">OrthoFinder</option>
|
|
|
120 <option value="orthomcl">OrthoMCL</option>
|
|
|
121 </param>
|
|
13
|
122 <conditional name="save_hmmscan_log_cond">
|
|
|
123 <param name="classifier" type="select" label="Protein classification method">
|
|
|
124 <option value="blastp" selected="true">blastp</option>
|
|
|
125 <option value="hmmscan">HMMScan</option>
|
|
|
126 <option value="both">Both blastp and HMMScan</option>
|
|
|
127 </param>
|
|
|
128 <when value="blastp" />
|
|
|
129 <when value="hmmscan">
|
|
|
130 <param name="save_hmmscan_log" type="select" label="Save hmmscan log?" help="Save the hmmscan log in an additional output dataset">
|
|
|
131 <option value="no" selected="true">No</option>
|
|
|
132 <option value="yes">Yes</option>
|
|
|
133 </param>
|
|
|
134 </when>
|
|
|
135 <when value="both">
|
|
|
136 <param name="save_hmmscan_log" type="select" label="Save hmmscan log?" help="Save the hmmscan log in an additional output dataset">
|
|
|
137 <option value="no" selected="true">No</option>
|
|
|
138 <option value="yes">Yes</option>
|
|
|
139 </param>
|
|
|
140 </when>
|
|
|
141 </conditional>
|
|
0
|
142 <conditional name="options_type">
|
|
|
143 <param name="options_type_selector" type="select" label="Options Configuration">
|
|
|
144 <option value="basic" selected="true">Basic</option>
|
|
|
145 <option value="advanced">Advanced</option>
|
|
|
146 </param>
|
|
|
147 <when value="basic" />
|
|
|
148 <when value="advanced">
|
|
51
|
149 <conditional name="specify_super_orthogroups_cond">
|
|
|
150 <param name="specify_super_orthogroups" type="select" label="Specify super orthogroups?" help="Secondary MCL clusters of orthogroups">
|
|
|
151 <option value="no" selected="true">No</option>
|
|
|
152 <option value="yes">Yes</option>
|
|
|
153 </param>
|
|
|
154 <when value="no"/>
|
|
|
155 <when value="yes">
|
|
97
|
156 <param name="super_orthogroups" type="select" label="Super orthogroups clustering specification">
|
|
51
|
157 <option value="min_evalue" selected="true">Minimum e-value</option>
|
|
|
158 <option value="avg_evalue">Average e-value</option>
|
|
|
159 </param>
|
|
|
160 </when>
|
|
|
161 </conditional>
|
|
44
|
162 <conditional name="specify_single_copy_cond">
|
|
|
163 <param name="specify_single_copy" type="select" label="Specify single copy orthogroup selection?">
|
|
|
164 <option value="no" selected="true">No</option>
|
|
|
165 <option value="yes">Yes</option>
|
|
0
|
166 </param>
|
|
44
|
167 <when value="no"/>
|
|
|
168 <when value="yes">
|
|
|
169 <conditional name="single_copy_cond">
|
|
73
|
170 <param name="single_copy" type="select" label="Select single copy orthogroup configuration option">
|
|
68
|
171 <option value="custom" selected="true">Single copy orthogroup custom configuration</option>
|
|
65
|
172 <option value="taxa">Minimum single copy taxa required in orthogroup</option>
|
|
44
|
173 </param>
|
|
70
|
174 <when value="custom">
|
|
|
175 <conditional name="single_copy_custom_cond">
|
|
73
|
176 <param name="single_copy_custom" type="select" label="Select single copy orthogroup custom configuration from the current history?" help="Select No to use the default configuration">
|
|
70
|
177 <option value="no" selected="true">No</option>
|
|
|
178 <option value="yes">Yes</option>
|
|
|
179 </param>
|
|
|
180 <when value="no"/>
|
|
|
181 <when value="yes">
|
|
73
|
182 <param name="single_copy_custom_config" format="txt" type="data" label="Single copy orthogroup custom configuration file"/>
|
|
70
|
183 </when>
|
|
|
184 </conditional>
|
|
|
185 </when>
|
|
44
|
186 <when value="taxa">
|
|
65
|
187 <param name="single_copy_taxa" type="integer" value="20" label="Minimum single copy taxa required in orthogroup"/>
|
|
|
188 <param name="taxa_present" type="integer" value="21" label="Minimum taxa required in single copy orthogroup"/>
|
|
44
|
189 </when>
|
|
|
190 </conditional>
|
|
0
|
191 </when>
|
|
|
192 </conditional>
|
|
|
193 <conditional name="create_orthogroup_cond">
|
|
|
194 <param name="create_orthogroup" type="select" label="Create orthogroup fasta files?">
|
|
|
195 <option value="no" selected="true">No</option>
|
|
|
196 <option value="yes">Yes</option>
|
|
|
197 </param>
|
|
|
198 <when value="no" />
|
|
|
199 <when value="yes">
|
|
17
|
200 <conditional name="create_corresponding_coding_sequences_cond">
|
|
|
201 <param name="create_corresponding_coding_sequences" type="select" label="Create corresponding coding sequences?">
|
|
|
202 <option value="no" selected="true">No</option>
|
|
|
203 <option value="yes">Yes</option>
|
|
|
204 </param>
|
|
|
205 <when value="no" />
|
|
|
206 <when value="yes">
|
|
|
207 <param name="coding_sequences" format="fasta" type="data" label="Corresponding coding sequences (CDS) fasta file"/>
|
|
|
208 </when>
|
|
|
209 </conditional>
|
|
0
|
210 </when>
|
|
|
211 </conditional>
|
|
|
212 </when>
|
|
|
213 </conditional>
|
|
|
214 </inputs>
|
|
|
215 <outputs>
|
|
15
|
216 <data name="hmmscan_log" format="txt" label="Protein classification hmmscan.log on ${on_string}">
|
|
|
217 <filter>save_hmmscan_log_cond['classifier'] in ['hmmscan', 'both'] and save_hmmscan_log_cond['save_hmmscan_log'] == 'yes'</filter>
|
|
|
218 </data>
|
|
63
|
219 <data name="output_ptortho" format="ptortho" label="Gene family clusters on ${on_string}">
|
|
33
|
220 <filter>options_type['options_type_selector'] == 'advanced' and options_type['create_orthogroup_cond']['create_orthogroup'] == 'yes' and options_type['create_orthogroup_cond']['create_corresponding_coding_sequences_cond']['create_corresponding_coding_sequences'] == 'no'</filter>
|
|
31
|
221 </data>
|
|
63
|
222 <data name="output_ptorthocs" format="ptorthocs" label="Gene family clusters and corresponding coding sequences on ${on_string}">
|
|
31
|
223 <filter>options_type['options_type_selector'] == 'advanced' and options_type['create_orthogroup_cond']['create_orthogroup'] == 'yes' and options_type['create_orthogroup_cond']['create_corresponding_coding_sequences_cond']['create_corresponding_coding_sequences'] == 'yes'</filter>
|
|
9
|
224 </data>
|
|
63
|
225 <data name="output_ptsco" format="tabular" label="Single copy orthogroups on ${on_string}">
|
|
53
|
226 <filter>options_type['options_type_selector'] == 'advanced' and options_type['create_orthogroup_cond']['create_orthogroup'] == 'yes' and options_type['specify_single_copy_cond']['specify_single_copy'] == 'yes'</filter>
|
|
63
|
227 <change_format>
|
|
|
228 <when input="options_type.create_orthogroup_cond.create_corresponding_coding_sequences_cond.create_corresponding_coding_sequences" value="no" format="ptortho" />
|
|
|
229 <when input="options_type.create_orthogroup_cond.create_corresponding_coding_sequences_cond.create_corresponding_coding_sequences" value="yes" format="ptorthocs" />
|
|
|
230 </change_format>
|
|
52
|
231 </data>
|
|
30
|
232 <collection name="orthos" type="list">
|
|
12
|
233 <discover_datasets pattern="__name__" directory="geneFamilyClassification_dir" visible="false" ext="tabular" />
|
|
0
|
234 </collection>
|
|
|
235 </outputs>
|
|
|
236 <tests>
|
|
|
237 <test>
|
|
30
|
238 <param name="input" value="transcripts.cleaned.nr.pep" ftype="fasta" />
|
|
|
239 <param name="scaffold" value="22Gv1.1"/>
|
|
|
240 <param name="method" value="orthomcl"/>
|
|
|
241 <param name="classifier" value="blastp"/>
|
|
0
|
242 <param name="dereplicate" value="yes"/>
|
|
|
243 <param name="min_length" value="200"/>
|
|
|
244 <output_collection name="orthos" type="list">
|
|
35
|
245 <element name="proteins.blastp.22Gv1.1" file="proteins.blastp.22Gv1.1" ftype="tabular"/>
|
|
|
246 <element name="proteins.blastp.22Gv1.1.bestOrthos" file="proteins.blastp.22Gv1.1.bestOrthos" ftype="tabular"/>
|
|
|
247 <element name="proteins.blastp.22Gv1.1.bestOrthos.summary" file="proteins.blastp.22Gv1.1.bestOrthos.summary" ftype="tabular"/>
|
|
0
|
248 </output_collection>
|
|
|
249 </test>
|
|
|
250 </tests>
|
|
|
251 <help>
|
|
|
252 This tool is one of the PlantTribes' collection of automated modular analysis pipelines that utilize objective classifications of
|
|
28
|
253 complete protein sequences from sequenced plant genomes to perform comparative evolutionary studies. This tool classifies gene
|
|
|
254 sequences into precomputed orthologous gene family clusters using either blastp (faster), HMMScan (slower but more sensitive
|
|
|
255 to remote homologs) or both (more exhaustive).
|
|
|
256
|
|
30
|
257 This tool accepts any of the following as input:
|
|
28
|
258
|
|
|
259 * the postprocessed assemblies produced by the **Postprocess de novo assembly transcripts into putative coding sequences** tool
|
|
|
260 * externally predicted coding sequences and their corresponding amino acid translations derived from a transcriptome assembly
|
|
|
261 * gene predictions from a sequenced genome
|
|
0
|
262
|
|
|
263 -----
|
|
|
264
|
|
|
265 **Options**
|
|
|
266
|
|
67
|
267 * **Orthogroups or gene families proteins scaffold** - PlantTribes scaffolds data installed into Galaxy by the PlantTribes Scaffolds Download Data Manager tool.
|
|
28
|
268 * **Protein clustering method** - One of GFam (domain architecture based clustering), OrthoFinder (broadly defined clusters) or OrthoMCL (narrowly defined clusters).
|
|
|
269 * **Protein classification method** - blastp (faster), HMMScan (slower but more sensative to the remote homologs) or both (more exhaustive).
|
|
|
270 * **Super Orthogroups** - Secondary MCL clusters of orthogroups.
|
|
73
|
271 * **Specify single copy orthogroup selection?** - Specify a single copy orthogroup custom configuration or the minimum single copy taxa required in the orthogroup.
|
|
|
272 * **Select single copy orthogroup custom configuration from the current history?** - If a custom configuration is chosen, the configuration can be selected from the current history or the default configuration can be used.
|
|
65
|
273 * **Minimum single copy taxa required in orthogroup** - Used with "Minimum single copy taxa required in orthogroup" configuration only.
|
|
|
274 * **Minimum taxa required in single copy orthogroup** - Used with "Minimum single copy taxa required in orthogroup" configuration only.
|
|
0
|
275 * **Corresponding coding sequences (CDS) fasta file** - Used only when selecting "Create orthogroup fasta files?".
|
|
|
276
|
|
|
277 </help>
|
|
|
278 <citations>
|
|
|
279 <citation type="bibtex">
|
|
|
280 @unpublished{None,
|
|
28
|
281 author = {Eric Wafula},
|
|
0
|
282 title = {None},
|
|
|
283 year = {None},
|
|
35
|
284 url = {https://github.com/dePamphilis/PlantTribes}
|
|
|
285 }</citation>
|
|
|
286 <citation type="doi">10.1186/1471-2105-10-421</citation>
|
|
|
287 <citation type="bibtex">
|
|
|
288 @unpublished{None,
|
|
|
289 author = {None},
|
|
|
290 title = {HMMER 3.1+ hmmscan search sequence(s) against a profile database},
|
|
|
291 year = {2013},
|
|
|
292 url = {http://hmmer.org/}
|
|
0
|
293 }</citation>
|
|
|
294 </citations>
|
|
|
295 </tool>
|
