Mercurial > repos > iuc > orthofinder_onlygroups

comparison orthofinder_only_groups.xml @ 5:974d8c28e586 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/orthofinder commit 8fb4e5b2821a83fda89ac616b32726714962717c"
author iuc
date Mon, 17 Aug 2020 12:59:19 -0400
parents 999060f051ac
children 0ff17c7a18cd
comparison
equal deleted inserted replaced
4:999060f051ac 5:974d8c28e586
1 <tool name="OrthoFinder OnlyGroups" id="orthofinder_onlygroups" version="2.2.6"> 1 <tool name="OrthoFinder" id="orthofinder_onlygroups" version="@TOOL_VERSION@">
2 <description>finds orthogroups in a set of proteomes</description> 2 <description>finds orthogroups in a set of proteomes</description>
3 <macros>
4 <import>macros.xml</import>
5 </macros>
3 <requirements> 6 <requirements>
4 <requirement type="package" version="2.2.6">orthofinder</requirement> 7 <requirement type="package" version="@TOOL_VERSION@">orthofinder</requirement>
5 <requirement type="package" version="2.34">util-linux</requirement> 8 <requirement type="package" version="2.34">util-linux</requirement>
6 </requirements> 9 </requirements>
7 <command detect_errors="exit_code"> 10 <command detect_errors="exit_code"><![CDATA[
8 <![CDATA[ 11 #import re
9 ## prepare inputs 12 ## prepare inputs
10 #if $init.start == "fasta": 13 #if $init.start == "fasta":
11 #set $infiles = "" 14 #set $infiles = ""
12 #for $input in $init.input_fasta 15 #for $input in $init.input_fasta
13 ln -s '$input' '${input.element_identifier}.fasta' && 16 #set $identifier=re.sub('[^\w\-\s\.]', '_', str($input.element_identifier))
14 #set $infiles = $infiles + str($input.element_identifier) + "," 17 ln -s '$input' '${identifier}.fasta' &&
18 #set $infiles = $infiles + str($identifier) + ","
15 #end for 19 #end for
16 #set $infiles = $infiles[:-1] 20 #set $infiles = $infiles[:-1]
17 #elif $init.start == "blast": 21 #elif $init.start == "blast":
18 #set $infilesbl = "" 22 #set $infilesbl = ""
19 #for $input in $init.input_blast_out 23 #for $input in $init.input_blast_out
20 ln -s '$input' '$input.element_identifier' && 24 #set $identifier=re.sub('[^\w\-\s\.]', '_', str($input.element_identifier))
21 #set $infilesbl = $infilesbl + str($input.element_identifier) + "," 25 ln -s '$input' '$identifier' &&
26 #set $infilesbl = $infilesbl + str($identifier) + ","
22 #end for 27 #end for
23 #set $infilesbl = $infilesbl[:-1] 28 #set $infilesbl = $infilesbl[:-1]
24 29
25 #set $infilesfa = "" 30 #set $infilesfa = ""
26 #for $input in $init.input_blast_fa 31 #for $input in $init.input_blast_fa
27 ln -s '$input' '$input.element_identifier' && 32 #set $identifier=re.sub('[^\w\-\s\.]', '_', str($input.element_identifier))
28 #set $infilesfa = $infilesfa + str($input.element_identifier) + "," 33 ln -s '$input' '$identifier' &&
34 #set $infilesfa = $infilesfa + str($identifier) + ","
29 #end for 35 #end for
30 #set $infilesfa = $infilesfa[:-1] 36 #set $infilesfa = $infilesfa[:-1]
31 37
32 ln -s $init.specIDs $init.specIDs.element_identifier && 38 #set $identifier=re.sub('[^\w\-\s\.]', '_', str($init.specIDs.element_identifier))
33 ln -s $init.seqIDs $init.seqIDs.element_identifier && 39 ln -s $init.specIDs $identifier &&
40
41 #set $identifier=re.sub('[^\w\-\s\.]', '_', str($init.seqIDs.element_identifier))
42 ln -s $init.seqIDs $identifier &&
34 #end if 43 #end if
35 44
36 ## start Orthofinder 45 ## start Orthofinder
37 orthofinder 46 orthofinder
38 #if $init.start == "fasta": 47 #if $init.start == "fasta":
40 -S $init.search.search_program 49 -S $init.search.search_program
41 #elif $init.start == "blast": 50 #elif $init.start == "blast":
42 -b . 51 -b .
43 #end if 52 #end if
44 53
45 -I $I -og -t \${GALAXY_SLOTS:-1} -a \${GALAXY_SLOTS:-1} && 54 -I $I
46 55
56 #if $trees.run_mode == "full":
57 -M '${trees.tree_method.method}'
58 #if $trees.tree_method.method == "msa":
59 -A '${trees.tree_method.msa_program}'
60 -T '${trees.tree_method.msa_tree_program}'
61 #end if
62 #else:
63 $trees.run_mode
64 #end if
65
66 -t \${GALAXY_SLOTS:-1} -a \${GALAXY_SLOTS:-1} &&
67
68 mv OrthoFinder/Results_* results
47 #if $init.start == "fasta": 69 #if $init.start == "fasta":
48 mv Results_* results
49 #if $init.search.search_program == "blast": 70 #if $init.search.search_program == "blast":
50 #if $init.search.keepblastout: 71 #if $init.search.keepblastout:
51 && mkdir -p results/WorkingDirectory/blast results/WorkingDirectory/fa && 72 && mkdir -p results/WorkingDirectory/blast results/WorkingDirectory/fa &&
52 mv results/WorkingDirectory/Blast* results/WorkingDirectory/blast/ && 73 mv results/WorkingDirectory/Blast* results/WorkingDirectory/blast/ &&
53 mv results/WorkingDirectory/*.fa results/WorkingDirectory/fa/ 74 mv results/WorkingDirectory/*.fa results/WorkingDirectory/fa/
54 #end if 75 #end if
55 #end if 76 #end if
56 #elif $init.start == "blast":
57 mkdir results &&
58 mv *.csv results/ &&
59 mv Orthogroups.txt results/
60 #end if 77 #end if
61 ]]> 78 ]]></command>
62 </command>
63 <inputs> 79 <inputs>
64 <!-- Control where Orthofinder starts --> 80 <!-- Control where Orthofinder starts -->
65 <conditional name="init"> 81 <conditional name="init">
66 <param name="start" type="select" label="Orthofinder starting point" help="OrthoFinder_OnlyGroups works in 2 steps. Choose 'From fasta proteomes' to run OrthoFinder_OnlyGroups from scratch and 'From blast results' if you have all the blast results from a previous OrthoFinder_OnlyGroups run."> 82 <param name="start" type="select" label="Orthofinder starting point" help="OrthoFinder can be run in 2 steps. Choose 'From fasta proteomes' to run OrthoFinder from scratch or 'From blast results' if you have all the blast results from a previous OrthoFinder run.">
67 <option value="fasta" selected="true">From fasta proteomes</option> 83 <option value="fasta" selected="true">From fasta proteomes</option>
68 <option value="blast">From blast results</option> 84 <option value="blast">From blast results</option>
69 </param> 85 </param>
70 86
71 <when value="fasta"> 87 <when value="fasta">
72 <param name="input_fasta" type="data" format="fasta" multiple="true" label="Select input fasta proteomes" help="One fasta file per species; species and sequences names in the results will remain the same than in the input files."/> 88 <param name="input_fasta" type="data" format="fasta" multiple="true" label="Select input fasta proteomes" help="One fasta file per species; species and sequences names in the results will remain the same than in the input files."/>
73 <conditional name="search"> 89 <conditional name="search">
74 <param name="search_program" type="select" label="Sequence search program" help="Choose between blast, blast_gz, diamond"> 90 <param name="search_program" type="select" label="Sequence search program">
75 <option value="blast" selected="true">blast</option> 91 <option value="diamond" selected="true">Diamond (faster)</option>
76 <option value="blast_gz">blast_gz</option> 92 <option value="blast">Blast</option>
77 <option value="diamond">diamond</option> 93 <option value="blast_gz">Blast_gz - blast results gzipped</option>
78 </param> 94 </param>
79 <when value="blast"> 95 <when value="blast">
80 <param name="keepblastout" type="boolean" checked="true" label="Do you want to get the blast results?" help="Used to re-run OrthoFinder_OnlyGroups from pre-computed blast results"/> 96 <param name="keepblastout" type="boolean" checked="true" label="Do you want to get the blast results?" help="Used to re-run OrthoFinder from pre-computed blast results"/>
81 </when> 97 </when>
82 <when value="diamond"></when> 98 <when value="diamond"></when>
83 <when value="blast_gz"></when> 99 <when value="blast_gz"></when>
84 </conditional> 100 </conditional>
85 </when> 101 </when>
86 102
87 <when value="blast"> 103 <when value="blast">
88 <param name="input_blast_out" type="data_collection" collection_type="list" format="txt" label="Select the pre-computed blast files" help="blastX_Y.txt files from the blast output files of a previous OrthoFinder_OnlyGroups run." /> 104 <param name="input_blast_out" type="data_collection" collection_type="list" format="txt" label="Select the pre-computed blast files" help="blastX_Y.txt files from the blast output files of a previous OrthoFinder run." />
89 <param name="input_blast_fa" type="data_collection" collection_type="list" format="fasta" label="Select the fasta files" help="= SpeciesX.fa files from the blast output files of a previous OrthoFinder_OnlyGroups run." /> 105 <param name="input_blast_fa" type="data_collection" collection_type="list" format="fasta" label="Select the fasta files" help="= SpeciesX.fa files from the blast output files of a previous OrthoFinder run." />
90 <param name="specIDs" type="data" format="txt" label="Select the SpeciesIds file" help="SpeciesIDs.txt file from the blast output files of a previous OrthoFinder_OnlyGroups run."/> 106 <param name="specIDs" type="data" format="txt" label="Select the SpeciesIds file" help="SpeciesIDs.txt file from the blast output files of a previous OrthoFinder run."/>
91 <param name="seqIDs" type="data" format="txt" label="Select the SequencesIds file" help="SequencesIDs.txt file from the blast output files of a previous OrthoFinder_OnlyGroups run."/> 107 <param name="seqIDs" type="data" format="txt" label="Select the SequencesIds file" help="SequencesIDs.txt file from the blast output files of a previous OrthoFinder run."/>
92 </when> 108 </when>
93 </conditional> 109 </conditional>
110
111 <conditional name="trees">
112 <param name="run_mode" type="select" label="Orthofinder run mode">
113 <option value="full" selected="true">Full run (including gene trees)</option>
114 <option value="-og">Stop after inferring orthogroups (no gene trees)</option>
115 </param>
116
117 <when value="full">
118 <conditional name="tree_method">
119 <param name="method" type="select" label="Method for gene tree inference">
120 <option value="dendroblast" selected="true">Dendroblast (faster)</option>
121 <option value="msa">MSA (Multiple Sequence Aligments)</option>
122 </param>
123
124 <when value="msa">
125 <param name="msa_program" type="select" label="MSA program">
126 <option value="mafft" selected="true">Mafft</option>
127 <option value="muscle">Muscle</option>
128 </param>
129 <param name="msa_tree_program" type="select" label="Tree inference method">
130 <option value="fasttree" selected="true">FastTree (recommended)</option>
131 <option value="raxml">raxml</option>
132 <option value="raxml-ng">raxml-ng</option>
133 <option value="iqtree">iqtree</option>
134 </param>
135 </when>
136
137 <when value="dendroblast"/>
138 </conditional>
139 </when>
140
141 <when value="-og"/>
142 </conditional>
143
94 <param argument="-I" type="float" value="1.5" label="Inflation parameter" help="Modify inflation parameter for MCL. Not recommended." /> 144 <param argument="-I" type="float" value="1.5" label="Inflation parameter" help="Modify inflation parameter for MCL. Not recommended." />
145 <param name="output_duplications" type="boolean" checked="false" label="Generate output about gene duplication events"/>
95 </inputs> 146 </inputs>
96 <outputs> 147 <outputs>
97 <!-- Orthogroups results --> 148 <!-- Orthogroups results -->
98 <data format="txt" name="orthogroups1" label="Orthogroups.txt" from_work_dir="results/Orthogroups.txt" /> 149 <data format="txt" name="orthogroups1" label="OrthoFinder on ${on_string}: orthogroups (txt)" from_work_dir="results/Orthogroups/Orthogroups.txt" />
99 <data format="csv" name="orthogroups2" label="Orthogroups.csv" from_work_dir="results/Orthogroups.csv" /> 150 <data format="tsv" name="orthogroups2" label="OrthoFinder on ${on_string}: orthogroups (tsv)" from_work_dir="results/Orthogroups/Orthogroups.tsv" />
100 <data format="csv" name="specs_overlap" label="Orthogroups_SpeciesOverlaps.csv" from_work_dir="results/Orthogroups_SpeciesOverlaps.csv" /> 151 <data format="tsv" name="specs_overlap" label="OrthoFinder on ${on_string}: species overlaps" from_work_dir="results/Comparative_Genomics_Statistics/Orthogroups_SpeciesOverlaps.tsv" />
101 <data format="csv" name="unassigned_genes" label="Orthogroups_UnassignedGenes.csv" from_work_dir="results/Orthogroups_UnassignedGenes.csv" /> 152 <data format="tsv" name="unassigned_genes" label="OrthoFinder on ${on_string}: unassigned genes" from_work_dir="results/Orthogroups/Orthogroups_UnassignedGenes.tsv" />
102 <data format="csv" name="stat_overall" label="Statistics_Overall.csv" from_work_dir="results/Statistics_Overall.csv" /> 153 <data format="tsv" name="stat_overall" label="OrthoFinder on ${on_string}: overall comparative genomics statistics" from_work_dir="results/Comparative_Genomics_Statistics/Statistics_Overall.tsv" />
103 <data format="csv" name="stat_specs" label="Statistics_PerSpecies.csv" from_work_dir="results/Statistics_PerSpecies.csv" /> 154 <data format="tsv" name="stat_specs" label="OrthoFinder on ${on_string}: per species comparative genomics statistics" from_work_dir="results/Comparative_Genomics_Statistics/Statistics_PerSpecies.tsv" />
104 155
105 <!-- working directory : blast outputs--> 156 <!-- working directory : blast outputs-->
106 <collection name="wdblast" type="list" label="Blast_outputs"> 157 <collection name="wdblast" type="list" label="OrthoFinder on ${on_string}: blast outputs">
107 <discover_datasets pattern="__name_and_ext__" directory="results/WorkingDirectory/blast" /> 158 <discover_datasets pattern="__name_and_ext__" directory="results/WorkingDirectory/blast" />
108 <filter>init['start'] == "fasta" and init['search']['search_program'] == "blast" and init['search']['keepblastout']</filter> 159 <filter>init['start'] == "fasta" and init['search']['search_program'] == "blast" and init['search']['keepblastout']</filter>
109 </collection> 160 </collection>
110 <collection name="wdfasta" type="list" label="Fasta_from_blast" > 161 <collection name="wdfasta" type="list" label="OrthoFinder on ${on_string}: fasta from blast" >
111 <discover_datasets pattern="(?P&lt;designation&gt;.+)\.fa" directory="results/WorkingDirectory/fa" format="fasta" /> 162 <discover_datasets pattern="(?P&lt;designation&gt;.+)\.fa" directory="results/WorkingDirectory/fa" format="fasta" />
112 <filter>init['start'] == "fasta" and init['search']['search_program'] == "blast" and init['search']['keepblastout']</filter> 163 <filter>init['start'] == "fasta" and init['search']['search_program'] == "blast" and init['search']['keepblastout']</filter>
113 </collection> 164 </collection>
114 <collection name="genetrees" type="list" label="Gene trees"> 165 <data format="txt" name="SpeciesIDs" label="OrthoFinder on ${on_string}: SpeciesIDs" from_work_dir="results/WorkingDirectory/SpeciesIDs.txt" >
115 <discover_datasets pattern="(?P&lt;designation&gt;.+)\.txt" directory="results/WorkingDirectory/trees" format="nhx" />
116 <filter>dogenetrees</filter>
117 </collection>
118 <data format="txt" name="SpeciesIDs" label="SpeciesIDs.txt" from_work_dir="results/WorkingDirectory/SpeciesIDs.txt" >
119 <filter>init['start'] == "fasta" and init['search']['search_program'] == "blast" and init['search']['keepblastout']"</filter> 166 <filter>init['start'] == "fasta" and init['search']['search_program'] == "blast" and init['search']['keepblastout']"</filter>
120 </data> 167 </data>
121 <data format="txt" name="SequenceIDs" label="SequencesIDs.txt" from_work_dir="results/WorkingDirectory/SequenceIDs.txt" > 168 <data format="txt" name="SequenceIDs" label="OrthoFinder on ${on_string}: SequencesIDs" from_work_dir="results/WorkingDirectory/SequenceIDs.txt" >
122 <filter>init['start'] == "fasta" and init['search']['search_program'] == "blast" and init['search']['keepblastout']</filter> 169 <filter>init['start'] == "fasta" and init['search']['search_program'] == "blast" and init['search']['keepblastout']</filter>
123 </data> 170 </data>
171
172 <data format="newick" name="species_tree" label="OrthoFinder on ${on_string}: species tree" from_work_dir="results/Species_Tree/SpeciesTree_rooted.txt">
173 <filter>trees['run_mode'] == "full"</filter>
174 </data>
175 <data format="newick" name="species_tree_label" label="OrthoFinder on ${on_string}: species tree with node labels" from_work_dir="results/Species_Tree/SpeciesTree_rooted_node_labels.txt">
176 <filter>trees['run_mode'] == "full"</filter>
177 </data>
178 <collection name="genetrees" type="list" label="OrthoFinder on ${on_string}: gene trees">
179 <discover_datasets pattern="(?P&lt;designation&gt;.+)\.txt" directory="results/Gene_Trees" format="newick" />
180 <filter>trees['run_mode'] == "full"</filter>
181 </collection>
182
183 <data format="newick" name="species_tree_duplications" label="OrthoFinder on ${on_string}: species tree with duplication events" from_work_dir="results/Gene_Duplication_Events/SpeciesTree_Gene_Duplications_0.5_Support.txt">
184 <filter>trees['run_mode'] == "full" and output_duplications</filter>
185 </data>
186 <data format="tsv" name="duplications" label="OrthoFinder on ${on_string}: duplication events" from_work_dir="results/Gene_Duplication_Events/Duplications.tsv">
187 <filter>trees['run_mode'] == "full" and output_duplications</filter>
188 </data>
189 <data format="tsv" name="duplications_per_orthogroup" label="OrthoFinder on ${on_string}: duplications per orthogroup" from_work_dir="results/Comparative_Genomics_Statistics/Duplications_per_Orthogroup.tsv">
190 <filter>trees['run_mode'] == "full" and output_duplications</filter>
191 </data>
192 <data format="tsv" name="duplications_per_species_tree_node" label="OrthoFinder on ${on_string}: duplications per species tree node" from_work_dir="results/Comparative_Genomics_Statistics/Duplications_per_Species_Tree_Node.tsv">
193 <filter>trees['run_mode'] == "full" and output_duplications</filter>
194 </data>
195 <collection name="resolved_trees" type="list" label="OrthoFinder on ${on_string}: resolved gene trees">
196 <discover_datasets pattern="(?P&lt;designation&gt;.+)\.txt" directory="results/Resolved_Gene_Trees" format="newick" />
197 <filter>trees['run_mode'] == "full" and output_duplications</filter>
198 </collection>
124 </outputs> 199 </outputs>
125 <tests> 200 <tests>
126 <!-- test orthofinder -f . -og when input files has no extension fasta/faa/fa ... --> 201 <!-- no trees + diamond + input files have no extension fasta/faa/fa -->
127 <test> 202 <test>
128 <conditional name="init"> 203 <conditional name="init">
129 <param name="start" value="fasta" /> 204 <param name="start" value="fasta" />
130 <param name="input_fasta" ftype="fasta" value="inputs/no_fa_ext/AcAcaud_trinity,inputs/no_fa_ext/AmAmphi_trinity,inputs/no_fa_ext/ApApomp_trinity,inputs/no_fa_ext/AsAsp1_trinity" /> 205 <param name="input_fasta" ftype="fasta" value="inputs/no_fa_ext/AcAcaud_trinity,inputs/no_fa_ext/AmAmphi_trinity,inputs/no_fa_ext/ApApomp_trinity,inputs/no_fa_ext/AsAsp1_trinity" />
131 <conditional name="search"> 206 <conditional name="search">
132 <param name="search_program" value="blast"/> 207 <param name="search_program" value="diamond"/>
133 <param name="keepblastout" value="false" /> 208 </conditional>
134 </conditional> 209 </conditional>
210 <conditional name="trees">
211 <param name="run_mode" value="-og" />
135 </conditional> 212 </conditional>
136 <param name="inflation" value="1.5" /> 213 <param name="inflation" value="1.5" />
137 <output name="specs_overlap"> 214 <output name="specs_overlap">
138 <assert_contents> 215 <assert_contents>
139 <has_text text="AcAcaud_trinity"/> 216 <has_text text="AcAcaud_trinity"/>
177 <has_text text="Number of genes"/> 254 <has_text text="Number of genes"/>
178 <has_n_columns n="5"/> 255 <has_n_columns n="5"/>
179 </assert_contents> 256 </assert_contents>
180 </output> 257 </output>
181 </test> 258 </test>
182 <!-- test orthofinder -f -og --> 259 <!-- no trees + diamond + input files have extension fasta/faa/fa -->
183 <test> 260 <test>
184 <conditional name="init"> 261 <conditional name="init">
185 <param name="start" value="fasta" /> 262 <param name="start" value="fasta" />
186 <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" /> 263 <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
187 <conditional name="search"> 264 <conditional name="search">
188 <param name="search_program" value="blast"/> 265 <param name="search_program" value="diamond"/>
189 <param name="keepblastout" value="false" /> 266 </conditional>
190 </conditional> 267 </conditional>
268 <conditional name="trees">
269 <param name="run_mode" value="-og" />
191 </conditional> 270 </conditional>
192 <param name="inflation" value="1.5" /> 271 <param name="inflation" value="1.5" />
193 <output name="specs_overlap"> 272 <output name="specs_overlap">
194 <assert_contents> 273 <assert_contents>
195 <has_text text="Mycoplasma_agalactiae"/> 274 <has_text text="Mycoplasma_agalactiae"/>
233 <has_text text="Number of genes"/> 312 <has_text text="Number of genes"/>
234 <has_n_columns n="5"/> 313 <has_n_columns n="5"/>
235 </assert_contents> 314 </assert_contents>
236 </output> 315 </output>
237 </test> 316 </test>
238 317 <!-- no trees + diamond + input files have extension fasta/faa/fa + keep blast out -->
239 <test> 318 <test>
240 <conditional name="init"> 319 <conditional name="init">
241 <param name="start" value="fasta" /> 320 <param name="start" value="fasta" />
242 <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" /> 321 <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
243 <conditional name="search"> 322 <conditional name="search">
244 <param name="search_program" value="blast"/> 323 <param name="search_program" value="blast"/>
245 <param name="keepblastout" value="true" /> 324 <param name="keepblastout" value="true" />
246 </conditional> 325 </conditional>
326 </conditional>
327 <conditional name="trees">
328 <param name="run_mode" value="-og" />
247 </conditional> 329 </conditional>
248 <param name="inflation" value="1.5" /> 330 <param name="inflation" value="1.5" />
249 <output name="specs_overlap"> 331 <output name="specs_overlap">
250 <assert_contents> 332 <assert_contents>
251 <has_text text="Mycoplasma_agalactiae"/> 333 <has_text text="Mycoplasma_agalactiae"/>
292 </output> 374 </output>
293 <output name="SequenceIDs" value="inputs/blastids/SequenceIDs.txt" /> 375 <output name="SequenceIDs" value="inputs/blastids/SequenceIDs.txt" />
294 <output_collection name="wdfasta" type="list" count="4"/> 376 <output_collection name="wdfasta" type="list" count="4"/>
295 <output_collection name="wdblast" type="list" count="16"/> 377 <output_collection name="wdblast" type="list" count="16"/>
296 </test> 378 </test>
297 379 <!-- no trees + blast + input files have extension fasta/faa/fa-->
298 <!-- test -S diamond -->
299 <test> 380 <test>
300 <conditional name="init"> 381 <conditional name="init">
301 <param name="start" value="fasta" /> 382 <param name="start" value="fasta" />
302 <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" /> 383 <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
303 <conditional name="search"> 384 <conditional name="search">
304 <param name="search_program" value="diamond"/> 385 <param name="search_program" value="blast"/>
305 </conditional> 386 <param name="keepblastout" value="false" />
387 </conditional>
388 </conditional>
389 <conditional name="trees">
390 <param name="run_mode" value="-og" />
306 </conditional> 391 </conditional>
307 <param name="inflation" value="1.5" /> 392 <param name="inflation" value="1.5" />
308 <output name="specs_overlap"> 393 <output name="specs_overlap">
309 <assert_contents> 394 <assert_contents>
310 <has_text text="Mycoplasma_agalactiae"/> 395 <has_text text="Mycoplasma_agalactiae"/>
348 <has_text text="Number of genes"/> 433 <has_text text="Number of genes"/>
349 <has_n_columns n="5"/> 434 <has_n_columns n="5"/>
350 </assert_contents> 435 </assert_contents>
351 </output> 436 </output>
352 </test> 437 </test>
353 438 <!-- no trees + uploaded blast-->
354 <!-- test orthofinder -b -og -->
355 <test> 439 <test>
356 <conditional name="init"> 440 <conditional name="init">
357 <param name="start" value="blast" /> 441 <param name="start" value="blast" />
358 <param name="input_blast_out"> 442 <param name="input_blast_out">
359 <collection type="list"> 443 <collection type="list">
384 </collection> 468 </collection>
385 </param> 469 </param>
386 <param name="specIDs" ftype="txt" value="inputs/blastids/SpeciesIDs.txt"/> 470 <param name="specIDs" ftype="txt" value="inputs/blastids/SpeciesIDs.txt"/>
387 <param name="seqIDs" ftype="txt" value="inputs/blastids/SequenceIDs.txt"/> 471 <param name="seqIDs" ftype="txt" value="inputs/blastids/SequenceIDs.txt"/>
388 </conditional> 472 </conditional>
473 <conditional name="trees">
474 <param name="run_mode" value="-og" />
475 </conditional>
389 <param name="inflation" value="1.5" /> 476 <param name="inflation" value="1.5" />
390 <output name="orthogroups1" value="results_fromblast/Orthogroups.txt"/> 477 <output name="orthogroups1" value="results_fromblast/Orthogroups.txt"/>
391 <output name="orthogroups2" value="results_fromblast/Orthogroups.csv"/> 478 <output name="orthogroups2" value="results_fromblast/Orthogroups.tsv"/>
392 <output name="specs_overlap" value="results_fromblast/Orthogroups_SpeciesOverlaps.csv"/> 479 <output name="specs_overlap" value="results_fromblast/Orthogroups_SpeciesOverlaps.tsv"/>
393 <output name="unassigned_genes" value="results_fromblast/Orthogroups_UnassignedGenes.csv"/> 480 <output name="unassigned_genes" value="results_fromblast/Orthogroups_UnassignedGenes.tsv"/>
394 <output name="stat_overall" value="results_fromblast/Statistics_Overall.csv" lines_diff="2"/> 481 <output name="stat_overall" value="results_fromblast/Statistics_Overall.tsv" lines_diff="2"/>
395 <output name="stat_specs" value="results_fromblast/Statistics_PerSpecies.csv"/> 482 <output name="stat_specs" value="results_fromblast/Statistics_PerSpecies.tsv"/>
483 </test>
484 <!-- full mode + diamond + input files have extension fasta/faa/fa -->
485 <test>
486 <conditional name="init">
487 <param name="start" value="fasta" />
488 <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
489 <conditional name="search">
490 <param name="search_program" value="diamond"/>
491 </conditional>
492 </conditional>
493 <conditional name="trees">
494 <param name="run_mode" value="full" />
495 </conditional>
496 <param name="inflation" value="1.5" />
497 <output name="specs_overlap">
498 <assert_contents>
499 <has_text text="Mycoplasma_agalactiae"/>
500 <has_text text="Mycoplasma_gallisepticum"/>
501 <has_text text="Mycoplasma_genitalium"/>
502 <has_text text="Mycoplasma_hyopneumoniae"/>
503 <has_n_columns n="5"/>
504 </assert_contents>
505 </output>
506 <output name="unassigned_genes">
507 <assert_contents>
508 <has_text text="Mycoplasma_agalactiae"/>
509 <has_text text="Mycoplasma_gallisepticum"/>
510 <has_text text="Mycoplasma_genitalium"/>
511 <has_text text="Mycoplasma_hyopneumoniae"/>
512 <has_n_columns n="5"/>
513 </assert_contents>
514 </output>
515 <output name="stat_overall">
516 <assert_contents>
517 <has_text text="Number of genes in orthogroups"/>
518 <has_text text="Number of unassigned genes"/>
519 <has_text text="Percentage of orthogroups"/>
520 <has_text text="Number of orthogroups"/>
521 <has_text text="Number of genes"/>
522 <has_text text="G50 (assigned genes)"/>
523 <has_text text="G50 (all genes)"/>
524 <has_text text="O50 (assigned genes)"/>
525 <has_text text="O50 (all genes)"/>
526 </assert_contents>
527 </output>
528 <output name="stat_specs">
529 <assert_contents>
530 <has_text text="Mycoplasma_agalactiae"/>
531 <has_text text="Mycoplasma_gallisepticum"/>
532 <has_text text="Mycoplasma_genitalium"/>
533 <has_text text="Mycoplasma_hyopneumoniae"/>
534 <has_text text="Number of genes per-species in orthogroup"/>
535 <has_text text="Percentage of orthogroups"/>
536 <has_text text="Number of orthogroups"/>
537 <has_text text="Number of genes"/>
538 <has_n_columns n="5"/>
539 </assert_contents>
540 </output>
541 <output name="species_tree" value="results/SpeciesTree_rooted.txt" compare="sim_size"/>
542 <output_collection name="genetrees" type="list" count="0"/>
543 <output_collection name="resolved_trees" type="list" count="0"/>
544 </test>
545 <!-- full mode + diamond + input files have extension fasta/faa/fa + duplications -->
546 <test>
547 <conditional name="init">
548 <param name="start" value="fasta" />
549 <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
550 <conditional name="search">
551 <param name="search_program" value="diamond"/>
552 </conditional>
553 </conditional>
554 <conditional name="trees">
555 <param name="run_mode" value="full" />
556 </conditional>
557 <param name="inflation" value="1.5" />
558 <param name="output_duplications" value="true" />
559 <output name="specs_overlap">
560 <assert_contents>
561 <has_text text="Mycoplasma_agalactiae"/>
562 <has_text text="Mycoplasma_gallisepticum"/>
563 <has_text text="Mycoplasma_genitalium"/>
564 <has_text text="Mycoplasma_hyopneumoniae"/>
565 <has_n_columns n="5"/>
566 </assert_contents>
567 </output>
568 <output name="unassigned_genes">
569 <assert_contents>
570 <has_text text="Mycoplasma_agalactiae"/>
571 <has_text text="Mycoplasma_gallisepticum"/>
572 <has_text text="Mycoplasma_genitalium"/>
573 <has_text text="Mycoplasma_hyopneumoniae"/>
574 <has_n_columns n="5"/>
575 </assert_contents>
576 </output>
577 <output name="stat_overall">
578 <assert_contents>
579 <has_text text="Number of genes in orthogroups"/>
580 <has_text text="Number of unassigned genes"/>
581 <has_text text="Percentage of orthogroups"/>
582 <has_text text="Number of orthogroups"/>
583 <has_text text="Number of genes"/>
584 <has_text text="G50 (assigned genes)"/>
585 <has_text text="G50 (all genes)"/>
586 <has_text text="O50 (assigned genes)"/>
587 <has_text text="O50 (all genes)"/>
588 </assert_contents>
589 </output>
590 <output name="stat_specs">
591 <assert_contents>
592 <has_text text="Mycoplasma_agalactiae"/>
593 <has_text text="Mycoplasma_gallisepticum"/>
594 <has_text text="Mycoplasma_genitalium"/>
595 <has_text text="Mycoplasma_hyopneumoniae"/>
596 <has_text text="Number of genes per-species in orthogroup"/>
597 <has_text text="Percentage of orthogroups"/>
598 <has_text text="Number of orthogroups"/>
599 <has_text text="Number of genes"/>
600 <has_n_columns n="5"/>
601 </assert_contents>
602 </output>
603 <output name="species_tree" value="results/SpeciesTree_rooted.txt" compare="sim_size"/>
604 <output name="species_tree_duplications" value="results/SpeciesTree_Gene_Duplications_0.5_Support.txt" compare="sim_size"/>
605 <output name="duplications" value="results/Duplications.tsv" compare="sim_size"/>
606 <output name="duplications_per_orthogroup" value="results/Duplications_per_Orthogroup.tsv" compare="sim_size"/>
607 <output name="duplications_per_species_tree_node" value="results/Duplications_per_Species_Tree_Node.tsv" compare="sim_size"/>
608 <output_collection name="genetrees" type="list" count="325"/>
609 <output_collection name="resolved_trees" type="list" count="325"/>
610 </test>
611 <!-- trees + diamond + input files have no extension fasta/faa/fa + msa -->
612 <test>
613 <conditional name="init">
614 <param name="start" value="fasta" />
615 <param name="input_fasta" ftype="fasta" value="inputs/no_fa_ext/AcAcaud_trinity,inputs/no_fa_ext/AmAmphi_trinity,inputs/no_fa_ext/ApApomp_trinity,inputs/no_fa_ext/AsAsp1_trinity" />
616 <conditional name="search">
617 <param name="search_program" value="diamond"/>
618 </conditional>
619 </conditional>
620 <conditional name="trees">
621 <param name="run_mode" value="-og" />
622 <conditional name="tree_method">
623 <param name="method" value="msa" />
624 <param name="msa_program" value="muscle" />
625 <param name="msa_tree_program" value="raxml" />
626 </conditional>
627 </conditional>
628 <param name="inflation" value="1.5" />
629 <output name="specs_overlap">
630 <assert_contents>
631 <has_text text="AcAcaud_trinity"/>
632 <has_text text="AmAmphi_trinity"/>
633 <has_text text="ApApomp_trinity"/>
634 <has_text text="AsAsp1_trinity"/>
635 <has_n_columns n="5"/>
636 </assert_contents>
637 </output>
638 <output name="unassigned_genes">
639 <assert_contents>
640 <has_text text="AcAcaud_trinity"/>
641 <has_text text="AmAmphi_trinity"/>
642 <has_text text="ApApomp_trinity"/>
643 <has_text text="AsAsp1_trinity"/>
644 <has_n_columns n="5"/>
645 </assert_contents>
646 </output>
647 <output name="stat_overall">
648 <assert_contents>
649 <has_text text="Number of genes in orthogroups"/>
650 <has_text text="Number of unassigned genes"/>
651 <has_text text="Percentage of orthogroups"/>
652 <has_text text="Number of orthogroups"/>
653 <has_text text="Number of genes"/>
654 <has_text text="G50 (assigned genes)"/>
655 <has_text text="G50 (all genes)"/>
656 <has_text text="O50 (assigned genes)"/>
657 <has_text text="O50 (all genes)"/>
658 </assert_contents>
659 </output>
660 <output name="stat_specs">
661 <assert_contents>
662 <has_text text="AcAcaud_trinity"/>
663 <has_text text="AmAmphi_trinity"/>
664 <has_text text="ApApomp_trinity"/>
665 <has_text text="AsAsp1_trinity"/>
666 <has_text text="Number of genes per-species in orthogroup"/>
667 <has_text text="Percentage of orthogroups"/>
668 <has_text text="Number of orthogroups"/>
669 <has_text text="Number of genes"/>
670 <has_n_columns n="5"/>
671 </assert_contents>
672 </output>
673 </test>
674 <!-- trees + diamond + input files have no extension fasta/faa/fa + msa -->
675 <test>
676 <conditional name="init">
677 <param name="start" value="fasta" />
678 <param name="input_fasta" ftype="fasta" value="inputs/proteomes/Mycoplasma_agalactiae.faa,inputs/proteomes/Mycoplasma_gallisepticum.faa,inputs/proteomes/Mycoplasma_genitalium.faa,inputs/proteomes/Mycoplasma_hyopneumoniae.faa" />
679 <conditional name="search">
680 <param name="search_program" value="diamond"/>
681 </conditional>
682 </conditional>
683 <conditional name="trees">
684 <param name="run_mode" value="full" />
685 <conditional name="tree_method">
686 <param name="method" value="msa" />
687 <param name="msa_program" value="muscle" />
688 <param name="msa_tree_program" value="raxml" />
689 </conditional>
690 </conditional>
691 <param name="inflation" value="1.5" />
692 <output name="specs_overlap">
693 <assert_contents>
694 <has_text text="Mycoplasma_agalactiae"/>
695 <has_text text="Mycoplasma_gallisepticum"/>
696 <has_text text="Mycoplasma_genitalium"/>
697 <has_text text="Mycoplasma_hyopneumoniae"/>
698 <has_n_columns n="5"/>
699 </assert_contents>
700 </output>
701 <output name="unassigned_genes">
702 <assert_contents>
703 <has_text text="Mycoplasma_agalactiae"/>
704 <has_text text="Mycoplasma_gallisepticum"/>
705 <has_text text="Mycoplasma_genitalium"/>
706 <has_text text="Mycoplasma_hyopneumoniae"/>
707 <has_n_columns n="5"/>
708 </assert_contents>
709 </output>
710 <output name="stat_overall">
711 <assert_contents>
712 <has_text text="Number of genes in orthogroups"/>
713 <has_text text="Number of unassigned genes"/>
714 <has_text text="Percentage of orthogroups"/>
715 <has_text text="Number of orthogroups"/>
716 <has_text text="Number of genes"/>
717 <has_text text="G50 (assigned genes)"/>
718 <has_text text="G50 (all genes)"/>
719 <has_text text="O50 (assigned genes)"/>
720 <has_text text="O50 (all genes)"/>
721 </assert_contents>
722 </output>
723 <output name="stat_specs">
724 <assert_contents>
725 <has_text text="Mycoplasma_agalactiae"/>
726 <has_text text="Mycoplasma_gallisepticum"/>
727 <has_text text="Mycoplasma_genitalium"/>
728 <has_text text="Mycoplasma_hyopneumoniae"/>
729 <has_text text="Number of genes per-species in orthogroup"/>
730 <has_text text="Percentage of orthogroups"/>
731 <has_text text="Number of orthogroups"/>
732 <has_text text="Number of genes"/>
733 <has_n_columns n="5"/>
734 </assert_contents>
735 </output>
736 <assert_command>
737 <has_text text="-M 'msa'"/>
738 <has_text text="-A 'muscle'"/>
739 <has_text text="-T 'raxml'"/>
740 </assert_command>
396 </test> 741 </test>
397 </tests> 742 </tests>
398 <help> 743 <help>
399 ====================== 744 ======================
400 OrthoFinder OnlyGroups 745 OrthoFinder OnlyGroups
427 - Sequence search program : You can choose either blast, blast_gz, or diamond (diamond is faster) 772 - Sequence search program : You can choose either blast, blast_gz, or diamond (diamond is faster)
428 - Get the blast results : Check "Yes" if, while using blast as the sequence search program, you want to retrieve the blast output files 773 - Get the blast results : Check "Yes" if, while using blast as the sequence search program, you want to retrieve the blast output files
429 - Inflation : the inflation parameter; modify this parameter is not recommended. 774 - Inflation : the inflation parameter; modify this parameter is not recommended.
430 775
431 </help> 776 </help>
432 <citations> 777 <expand macro="citations"/>
433 <citation type="doi">10.1186/s13059-015-0721-2</citation>
434 </citations>
435 </tool> 778 </tool>