|
28
|
1 <tool id="plant_tribes_gene_family_classifier" name="Classify gene sequences" version="0.3">
|
|
|
2 <description>into precomputed orthologous gene family clusters</description>
|
|
0
|
3 <requirements>
|
|
28
|
4 <requirement type="package" version="0.3">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[
|
|
21
|
16 #if str($options_type.options_type_selector) == 'advanced':
|
|
17
|
17 #set create_orthogroup_cond = $options_type.create_orthogroup_cond
|
|
|
18 #set create_orthogroup = $create_orthogroup_cond.create_orthogroup
|
|
22
|
19
|
|
21
|
20 #if str($create_orthogroup) == 'yes':
|
|
|
21 #set create_corresponding_coding_sequences_cond = $create_orthogroup_cond.create_corresponding_coding_sequences_cond
|
|
|
22 #if str($create_corresponding_coding_sequences_cond.create_corresponding_coding_sequences) == 'yes':
|
|
|
23 #set create_corresponding_coding_sequences = True
|
|
|
24 #else:
|
|
|
25 #set create_corresponding_coding_sequences = False
|
|
22
|
26 #end if
|
|
|
27
|
|
21
|
28 #import os
|
|
|
29 #set create_ortho_sequences = True
|
|
|
30 #set orthogroups_fasta_src_dir = $os.path.join('geneFamilyClassification_dir', 'orthogroups_fasta')
|
|
|
31 #set dest_dir = $output.extra_files_path
|
|
23
|
32 mkdir -p $dest_dir &&
|
|
21
|
33 #else:
|
|
|
34 #set create_ortho_sequences = False
|
|
23
|
35 #set create_corresponding_coding_sequences = False
|
|
21
|
36 #end if
|
|
13
|
37 #else:
|
|
|
38 #set create_ortho_sequences = False
|
|
23
|
39 #set create_corresponding_coding_sequences = False
|
|
4
|
40 #end if
|
|
20
|
41
|
|
0
|
42 GeneFamilyClassifier
|
|
|
43 --proteins "$input"
|
|
30
|
44 --scaffold_dir "${GALAXY_DATA_INDEX_DIR}/plant_tribes/scaffolds"
|
|
0
|
45 --scaffold "$scaffold"
|
|
|
46 --method $method
|
|
13
|
47 --classifier $save_hmmscan_log_cond.classifier
|
|
30
|
48 --num_threads \${GALAXY_SLOTS:-4}
|
|
20
|
49
|
|
0
|
50 #if str($options_type.options_type_selector) == 'advanced':
|
|
|
51 --super_orthogroups $options_type.super_orthogroups
|
|
|
52 #if str($options_type.single_copy_cond) == 'taxa':
|
|
|
53 --single_copy_taxa $options_type.single_copy_cond.single_copy_taxa
|
|
|
54 --taxa_present $options_type.single_copy_cond.taxa_present
|
|
|
55 #end if
|
|
17
|
56 #if str($create_orthogroup) == 'yes':
|
|
0
|
57 --orthogroup_fasta
|
|
18
|
58 #if $create_corresponding_coding_sequences:
|
|
17
|
59 --coding_sequences "$create_corresponding_coding_sequences_cond.coding_sequences"
|
|
|
60 #end if
|
|
0
|
61 #end if
|
|
|
62 #end if
|
|
20
|
63
|
|
13
|
64 #if str($save_hmmscan_log_cond.classifier) == 'hmmscan' or str($save_hmmscan_log_cond.classifier) == 'both':
|
|
|
65 #if str($save_hmmscan_log_cond.save_hmmscan_log) == 'yes':
|
|
15
|
66 && mv geneFamilyClassification_dir/hmmscan.log $hmmscan_log
|
|
13
|
67 #else:
|
|
15
|
68 && rm geneFamilyClassification_dir/hmmscan.log
|
|
13
|
69 #end if
|
|
12
|
70 #end if
|
|
20
|
71
|
|
14
|
72 #if $create_ortho_sequences:
|
|
18
|
73 #if $create_corresponding_coding_sequences:
|
|
30
|
74 && echo "# Precomputed orthologous gene family clusters with corresponding coding sequences: `ls $orthogroups_fasta_src_dir | grep f | wc -l` files" > $output
|
|
18
|
75 #else:
|
|
30
|
76 && echo "# Precomputed orthologous gene family clusters: `ls $orthogroups_fasta_src_dir | grep f | wc -l` files" > $output
|
|
18
|
77 #end if
|
|
25
|
78 && ls -al $orthogroups_fasta_src_dir | grep f >> $output
|
|
26
|
79 && mv $orthogroups_fasta_src_dir/* $dest_dir || true
|
|
4
|
80 #end if
|
|
20
|
81
|
|
0
|
82 ]]>
|
|
|
83 </command>
|
|
|
84 <inputs>
|
|
|
85 <param name="input" format="fasta" type="data" label="Amino acids (proteins) sequences fasta file"/>
|
|
|
86 <param name="scaffold" type="select" label="Orthogroups or gene families proteins scaffold">
|
|
|
87 <options from_data_table="plant_tribes_scaffolds" />
|
|
|
88 <validator type="no_options" message="No PlantTribes scaffolds are available. Use the PlantTribes Scaffolds Download Data Manager tool to install and populate the PlantTribes scaffolds data table."/>
|
|
|
89 </param>
|
|
|
90 <param name="method" type="select" label="Protein clustering method">
|
|
|
91 <option value="gfam" selected="true">GFam</option>
|
|
|
92 <option value="orthofinder">OrthoFinder</option>
|
|
|
93 <option value="orthomcl">OrthoMCL</option>
|
|
|
94 </param>
|
|
13
|
95 <conditional name="save_hmmscan_log_cond">
|
|
|
96 <param name="classifier" type="select" label="Protein classification method">
|
|
|
97 <option value="blastp" selected="true">blastp</option>
|
|
|
98 <option value="hmmscan">HMMScan</option>
|
|
|
99 <option value="both">Both blastp and HMMScan</option>
|
|
|
100 </param>
|
|
|
101 <when value="blastp" />
|
|
|
102 <when value="hmmscan">
|
|
|
103 <param name="save_hmmscan_log" type="select" label="Save hmmscan log?" help="Save the hmmscan log in an additional output dataset">
|
|
|
104 <option value="no" selected="true">No</option>
|
|
|
105 <option value="yes">Yes</option>
|
|
|
106 </param>
|
|
|
107 </when>
|
|
|
108 <when value="both">
|
|
|
109 <param name="save_hmmscan_log" type="select" label="Save hmmscan log?" help="Save the hmmscan log in an additional output dataset">
|
|
|
110 <option value="no" selected="true">No</option>
|
|
|
111 <option value="yes">Yes</option>
|
|
|
112 </param>
|
|
|
113 </when>
|
|
|
114 </conditional>
|
|
0
|
115 <conditional name="options_type">
|
|
|
116 <param name="options_type_selector" type="select" label="Options Configuration">
|
|
|
117 <option value="basic" selected="true">Basic</option>
|
|
|
118 <option value="advanced">Advanced</option>
|
|
|
119 </param>
|
|
|
120 <when value="basic" />
|
|
|
121 <when value="advanced">
|
|
28
|
122 <param name="super_orthogroups" type="select" label="Super Orthogroups" help="Secondary MCL clusters of orthogroups">
|
|
0
|
123 <option value="min_evalue" selected="true">Minimum e-value</option>
|
|
|
124 <option value="avg_evalue">Average e-value</option>
|
|
|
125 </param>
|
|
|
126 <conditional name="single_copy_cond">
|
|
|
127 <param name="single_copy" type="select" label="Select single copy configuration">
|
|
|
128 <option value="custom" selected="true">Single copy orthogroup custom</option>
|
|
|
129 <option value="taxa">Minumum single copy taxa required in orthogroup</option>
|
|
|
130 </param>
|
|
|
131 <when value="custom" />
|
|
|
132 <when value="taxa">
|
|
|
133 <param name="single_copy_taxa" type="integer" value="20" label="Minumum single copy taxa required in orthogroup"/>
|
|
|
134 <param name="taxa_present" type="integer" value="21" label="Minumum taxa required in single copy orthogroup"/>
|
|
|
135 </when>
|
|
|
136 </conditional>
|
|
|
137 <conditional name="create_orthogroup_cond">
|
|
|
138 <param name="create_orthogroup" type="select" label="Create orthogroup fasta files?">
|
|
|
139 <option value="no" selected="true">No</option>
|
|
|
140 <option value="yes">Yes</option>
|
|
|
141 </param>
|
|
|
142 <when value="no" />
|
|
|
143 <when value="yes">
|
|
17
|
144 <conditional name="create_corresponding_coding_sequences_cond">
|
|
|
145 <param name="create_corresponding_coding_sequences" type="select" label="Create corresponding coding sequences?">
|
|
|
146 <option value="no" selected="true">No</option>
|
|
|
147 <option value="yes">Yes</option>
|
|
|
148 </param>
|
|
|
149 <when value="no" />
|
|
|
150 <when value="yes">
|
|
|
151 <param name="coding_sequences" format="fasta" type="data" label="Corresponding coding sequences (CDS) fasta file"/>
|
|
|
152 </when>
|
|
|
153 </conditional>
|
|
0
|
154 </when>
|
|
|
155 </conditional>
|
|
|
156 </when>
|
|
|
157 </conditional>
|
|
|
158 </inputs>
|
|
|
159 <outputs>
|
|
15
|
160 <data name="hmmscan_log" format="txt" label="Protein classification hmmscan.log on ${on_string}">
|
|
|
161 <filter>save_hmmscan_log_cond['classifier'] in ['hmmscan', 'both'] and save_hmmscan_log_cond['save_hmmscan_log'] == 'yes'</filter>
|
|
|
162 </data>
|
|
32
|
163 <data name="output" format="pgfc" label="Gene family clusters on ${on_string}">
|
|
31
|
164 <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>
|
|
|
165 </data>
|
|
32
|
166 <data name="output" format="pgfccs" label="Gene family clusters and corresponding coding sequences on ${on_string}">
|
|
31
|
167 <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
|
168 </data>
|
|
30
|
169 <collection name="orthos" type="list">
|
|
12
|
170 <discover_datasets pattern="__name__" directory="geneFamilyClassification_dir" visible="false" ext="tabular" />
|
|
0
|
171 </collection>
|
|
|
172 </outputs>
|
|
|
173 <tests>
|
|
|
174 <test>
|
|
30
|
175 <!-- Not sure how to test this since the tool requires scaffolds data which is extremely large and installed using a Data Manager -->
|
|
|
176 <param name="input" value="transcripts.cleaned.nr.pep" ftype="fasta" />
|
|
|
177 <param name="scaffold" value="22Gv1.1"/>
|
|
|
178 <param name="method" value="orthomcl"/>
|
|
|
179 <param name="classifier" value="blastp"/>
|
|
0
|
180 <param name="dereplicate" value="yes"/>
|
|
|
181 <param name="min_length" value="200"/>
|
|
|
182 <output_collection name="orthos" type="list">
|
|
30
|
183 <element name="proteins.blastp.22Gv1.1" file="proteins.blastp.22Gv1.1" ftype="tabular" compare="contains"/>
|
|
|
184 <element name="proteins.blastp.22Gv1.1.bestOrthos" file="proteins.blastp.22Gv1.1.bestOrthos" ftype="tabular" compare="contains"/>
|
|
|
185 <element name="proteins.blastp.22Gv1.1.bestOrthos.summary" file="proteins.blastp.22Gv1.1.bestOrthos.summary" ftype="tabular" compare="contains"/>
|
|
0
|
186 </output_collection>
|
|
|
187 </test>
|
|
|
188 </tests>
|
|
|
189 <help>
|
|
|
190 This tool is one of the PlantTribes' collection of automated modular analysis pipelines that utilize objective classifications of
|
|
28
|
191 complete protein sequences from sequenced plant genomes to perform comparative evolutionary studies. This tool classifies gene
|
|
|
192 sequences into precomputed orthologous gene family clusters using either blastp (faster), HMMScan (slower but more sensitive
|
|
|
193 to remote homologs) or both (more exhaustive).
|
|
|
194
|
|
30
|
195 This tool accepts any of the following as input:
|
|
28
|
196
|
|
|
197 * the postprocessed assemblies produced by the **Postprocess de novo assembly transcripts into putative coding sequences** tool
|
|
|
198 * externally predicted coding sequences and their corresponding amino acid translations derived from a transcriptome assembly
|
|
|
199 * gene predictions from a sequenced genome
|
|
0
|
200
|
|
|
201 -----
|
|
|
202
|
|
|
203 **Options**
|
|
|
204
|
|
|
205 * **Orthogroups or gene families proteins scaffold** - PlantTribes scaffolds data.
|
|
28
|
206 * **Protein clustering method** - One of GFam (domain architecture based clustering), OrthoFinder (broadly defined clusters) or OrthoMCL (narrowly defined clusters).
|
|
|
207 * **Protein classification method** - blastp (faster), HMMScan (slower but more sensative to the remote homologs) or both (more exhaustive).
|
|
|
208 * **Super Orthogroups** - Secondary MCL clusters of orthogroups.
|
|
0
|
209 * **Minumum single copy taxa required in orthogroup** - Used with "Minumum single copy taxa required in orthogroup" configuration only.
|
|
|
210 * **Minumum taxa required in single copy orthogroup** - Used with "Minumum single copy taxa required in orthogroup" configuration only.
|
|
|
211 * **Corresponding coding sequences (CDS) fasta file** - Used only when selecting "Create orthogroup fasta files?".
|
|
|
212
|
|
|
213 </help>
|
|
|
214 <citations>
|
|
|
215 <citation type="bibtex">
|
|
|
216 @unpublished{None,
|
|
28
|
217 author = {Eric Wafula},
|
|
0
|
218 title = {None},
|
|
|
219 year = {None},
|
|
|
220 eprint = {None},
|
|
|
221 url = {None}
|
|
|
222 }</citation>
|
|
|
223 </citations>
|
|
|
224 </tool>
|
