|
1
|
1 <tool id="plant_tribes_gene_family_classifier" name="GeneFamilyClassifier" version="1.0.0">
|
|
0
|
2 <description>pipeline</description>
|
|
|
3 <requirements>
|
|
|
4 <requirement type="package" version="0.2">plant_tribes_gene_family_classifier</requirement>
|
|
|
5 </requirements>
|
|
|
6 <stdio>
|
|
|
7 <!-- Anything other than zero is an error -->
|
|
|
8 <exit_code range="1:" />
|
|
|
9 <!-- In case the return code has not been set properly check stderr too -->
|
|
|
10 <regex match="Error:" />
|
|
|
11 <regex match="Exception:" />
|
|
|
12 </stdio>
|
|
|
13 <command>
|
|
|
14 <![CDATA[
|
|
8
|
15 #if str($options_type.options_type_selector) == 'advanced' and str($options_type.create_orthogroup_cond.create_orthogroup) == 'yes':
|
|
|
16 #import os
|
|
|
17 #set orthogroups_src_dir = $os.path.join('geneFamilyClassification_dir', 'orthogroups_fasta')
|
|
|
18 #set orthogroups_dest_dir = $options_type.create_orthogroup_cond.coding_sequences.extra_files_path
|
|
|
19 mkdir -p $orthogroups_dest_dir &&
|
|
4
|
20 #end if
|
|
0
|
21 GeneFamilyClassifier
|
|
|
22 --proteins "$input"
|
|
|
23 --scaffold "$scaffold"
|
|
|
24 --method $method
|
|
|
25 --classifier $classifier
|
|
|
26 #if str($options_type.options_type_selector) == 'advanced':
|
|
|
27 --super_orthogroups $options_type.super_orthogroups
|
|
|
28 #if str($options_type.single_copy_cond) == 'taxa':
|
|
|
29 --single_copy_taxa $options_type.single_copy_cond.single_copy_taxa
|
|
|
30 --taxa_present $options_type.single_copy_cond.taxa_present
|
|
|
31 #end if
|
|
3
|
32 #if str($options_type.create_orthogroup_cond.create_orthogroup) == 'yes':
|
|
0
|
33 --orthogroup_fasta
|
|
|
34 --coding_sequences "$options_type.create_orthogroup_cond.coding_sequences"
|
|
|
35 #end if
|
|
|
36 #end if
|
|
2
|
37 --scaffold_dir "${GALAXY_DATA_INDEX_DIR}/plant_tribes/scaffolds"
|
|
0
|
38 --num_threads \${GALAXY_SLOTS:-4}
|
|
8
|
39 #if str($options_type.options_type_selector) == 'advanced' and str($options_type.create_orthogroup_cond.create_orthogroup) == 'yes':
|
|
|
40 && mv $orthogroups_src_dir/* $orthogroups_dest_dir
|
|
4
|
41 #end if
|
|
0
|
42 ]]>
|
|
|
43 </command>
|
|
|
44 <inputs>
|
|
|
45 <param name="input" format="fasta" type="data" label="Amino acids (proteins) sequences fasta file"/>
|
|
|
46 <param name="scaffold" type="select" label="Orthogroups or gene families proteins scaffold">
|
|
|
47 <options from_data_table="plant_tribes_scaffolds" />
|
|
|
48 <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."/>
|
|
|
49 </param>
|
|
|
50 <param name="method" type="select" label="Protein clustering method">
|
|
|
51 <option value="gfam" selected="true">GFam</option>
|
|
|
52 <option value="orthofinder">OrthoFinder</option>
|
|
|
53 <option value="orthomcl">OrthoMCL</option>
|
|
|
54 </param>
|
|
|
55 <param name="classifier" type="select" label="Protein classification method">
|
|
|
56 <option value="blastp" selected="true">blastp</option>
|
|
|
57 <option value="hmmscan">HMMScan</option>
|
|
|
58 <option value="both">Both blastp and HMMScan</option>
|
|
|
59 </param>
|
|
|
60 <conditional name="options_type">
|
|
|
61 <param name="options_type_selector" type="select" label="Options Configuration">
|
|
|
62 <option value="basic" selected="true">Basic</option>
|
|
|
63 <option value="advanced">Advanced</option>
|
|
|
64 </param>
|
|
|
65 <when value="basic" />
|
|
|
66 <when value="advanced">
|
|
|
67 <param name="super_orthogroups" type="select" label="SuperOrthogroups MCL clustering" help="blastp e-value matrix between all pairs of orthogroups">
|
|
|
68 <option value="min_evalue" selected="true">Minimum e-value</option>
|
|
|
69 <option value="avg_evalue">Average e-value</option>
|
|
|
70 </param>
|
|
|
71 <conditional name="single_copy_cond">
|
|
|
72 <param name="single_copy" type="select" label="Select single copy configuration">
|
|
|
73 <option value="custom" selected="true">Single copy orthogroup custom</option>
|
|
|
74 <option value="taxa">Minumum single copy taxa required in orthogroup</option>
|
|
|
75 </param>
|
|
|
76 <when value="custom" />
|
|
|
77 <when value="taxa">
|
|
|
78 <param name="single_copy_taxa" type="integer" value="20" label="Minumum single copy taxa required in orthogroup"/>
|
|
|
79 <param name="taxa_present" type="integer" value="21" label="Minumum taxa required in single copy orthogroup"/>
|
|
|
80 </when>
|
|
|
81 </conditional>
|
|
|
82 <conditional name="create_orthogroup_cond">
|
|
|
83 <param name="create_orthogroup" type="select" label="Create orthogroup fasta files?">
|
|
|
84 <option value="no" selected="true">No</option>
|
|
|
85 <option value="yes">Yes</option>
|
|
|
86 </param>
|
|
|
87 <when value="no" />
|
|
|
88 <when value="yes">
|
|
|
89 <param name="coding_sequences" format="fasta" type="data" label="Corresponding coding sequences (CDS) fasta file"/>
|
|
|
90 </when>
|
|
|
91 </conditional>
|
|
|
92 </when>
|
|
|
93 </conditional>
|
|
|
94 </inputs>
|
|
|
95 <outputs>
|
|
|
96 <collection name="transcripts" type="list">
|
|
1
|
97 <discover_datasets pattern="__name__" directory="geneFamilyClassification_dir" visible="true" ext="fasta" />
|
|
0
|
98 </collection>
|
|
|
99 </outputs>
|
|
|
100 <tests>
|
|
|
101 <test>
|
|
|
102 <param name="input" value="" ftype="fasta" />
|
|
|
103 <param name="prediction_method" value="transdecoder"/>
|
|
|
104 <param name="target_gene_family_assembly" value="no"/>
|
|
|
105 <param name="strand_specific" value="yes"/>
|
|
|
106 <param name="dereplicate" value="yes"/>
|
|
|
107 <param name="min_length" value="200"/>
|
|
|
108 <output_collection name="orthos" type="list">
|
|
|
109
|
|
|
110 </output_collection>
|
|
|
111 </test>
|
|
|
112 </tests>
|
|
|
113 <help>
|
|
|
114 This tool is one of the PlantTribes' collection of automated modular analysis pipelines that utilize objective classifications of
|
|
|
115 complete protein sequences from sequenced plant genomes to perform comparative evolutionary studies. It performs Gene family
|
|
|
116 classification of the post processed de novo transcripts using either blastp (faster), HMMScan (slower but more sensitive to remote
|
|
|
117 homologs) or both (more exhaustive).
|
|
|
118
|
|
|
119 -----
|
|
|
120
|
|
|
121 **Options**
|
|
|
122
|
|
|
123 * **Orthogroups or gene families proteins scaffold** - PlantTribes scaffolds data.
|
|
|
124 * **Protein clustering method** - One of GFam, OrthoFinder or OrthoMCL.
|
|
|
125 * **Protein classification method** - One of blastp, HMMScan or both.
|
|
|
126 * **SuperOrthogroups MCL clustering** - blastp e-value matrix between all pairs of orthogroups.
|
|
|
127 * **Minumum single copy taxa required in orthogroup** - Used with "Minumum single copy taxa required in orthogroup" configuration only.
|
|
|
128 * **Minumum taxa required in single copy orthogroup** - Used with "Minumum single copy taxa required in orthogroup" configuration only.
|
|
|
129 * **Corresponding coding sequences (CDS) fasta file** - Used only when selecting "Create orthogroup fasta files?".
|
|
|
130
|
|
|
131 </help>
|
|
|
132 <citations>
|
|
|
133 <citation type="bibtex">
|
|
|
134 @unpublished{None,
|
|
|
135 author = {None},
|
|
|
136 title = {None},
|
|
|
137 year = {None},
|
|
|
138 eprint = {None},
|
|
|
139 url = {None}
|
|
|
140 }</citation>
|
|
|
141 </citations>
|
|
|
142 </tool>
|
