Repository 'busco'
hg clone https://toolshed.g2.bx.psu.edu/repos/iuc/busco

Changeset 2:382a4c19007f (2017-10-20)
Previous changeset 1:87983967544b (2017-01-24) Next changeset 3:f18f0b887c31 (2018-12-14)
Commit message:
planemo upload commit 2896dcfd180800d00ea413a59264ef8b11788b8e
modified:
busco.xml
test-data/busco.loc
test-data/genome_results/full_table
test-data/genome_results/missing_buscos_list
test-data/genome_results/short_summary
test-data/proteome_results/full_table
test-data/proteome_results/missing_buscos_list
test-data/proteome_results/short_summary
test-data/transcriptome_results/full_table
test-data/transcriptome_results/missing_buscos_list
test-data/transcriptome_results/short_summary
tool-data/busco.loc.sample
tool_data_table_conf.xml.sample
tool_data_table_conf.xml.test
added:
test-data/local.tar.gz
b
diff -r 87983967544b -r 382a4c19007f busco.xml
--- a/busco.xml Tue Jan 24 12:08:38 2017 -0500
+++ b/busco.xml Fri Oct 20 03:54:35 2017 -0400
[
b'@@ -1,22 +1,39 @@\n-<tool id="busco" name="Busco" version="2.0">\n+<tool id="busco" name="Busco" version="3.0.2">\n     <description>assess genome assembly and annotation completeness</description>\n-\n     <requirements>\n-        <requirement type="package" version="2.0">busco</requirement>\n+        <requirement type="package" version="3.0.2">busco</requirement>\n+        <requirement type="package" version="2.2.1">r-ggplot2</requirement>\n     </requirements>\n+    <command><![CDATA[\n+        export BUSCO_CONFIG_FILE=\'busco_config.ini\'\n \n-    <command><![CDATA[\n-        BUSCO.py\n+        &&\n+\n+        #if $adv.aug_prediction.augustus_mode == \'history\':\n+            ## Using an augustus model from history, we need to unzip it and let augustus find it\n+\n+            cp -r "\\$AUGUSTUS_CONFIG_PATH/" augustus_dir/ &&\n+\n+            mkdir -p \'augustus_dir/species/\' &&\n+\n+            tar -C \'augustus_dir/species/\' -xzvf \'${adv.aug_prediction.augustus_model}\' > /dev/null &&\n+\n+            export AUGUSTUS_CONFIG_PATH=`pwd`/augustus_dir/ &&\n+        #end if\n+\n+        run_BUSCO.py\n         --in \'${input}\'\n-        --lineage \'${lineage.fields.path}\'\n+        --lineage_path \'${lineage_path.fields.path}/${lineage_path.fields.value}\'\n         --mode \'${mode}\'\n         -o "busco_galaxy"\n         --cpu \\${GALAXY_SLOTS:-4}\n         --evalue ${adv.evalue}\n         ${adv.long}\n         --limit ${adv.limit}\n-        #if $adv.species:\n-            --species \'${adv.species}\'\n+        #if $adv.aug_prediction.augustus_mode == \'builtin\':\n+            --species \'${adv.aug_prediction.augustus_species}\'\n+        #else if $adv.aug_prediction.augustus_mode == \'history\':\n+            --species local\n         #end if\n         --tarzip\n     ]]></command>\n@@ -24,14 +41,13 @@\n     <inputs>\n \n         <param type="data" name="input" format="fasta" label="Sequences to analyse" help="genome, transcriptome or proteome" />\n-\n         <param argument="--mode" type="select" label="Mode">\n             <option value="geno">Genome</option>\n             <option value="tran">Transcriptome</option>\n             <option value="prot">Proteome</option>\n         </param>\n \n-        <param argument="--lineage" type="select" label="Lineage">\n+        <param argument="--lineage_path" type="select" label="Lineage">\n             <options from_data_table="busco">\n                 <filter type="sort_by" column="2" />\n                 <validator type="no_options" message="No indexes are available" />\n@@ -41,7 +57,118 @@\n         <section name="adv" title="Advanced Options" expanded="False">\n             <param argument="--evalue" type="float" value="0.01" label="E-value cutoff for BLAST searches."/>\n             <param argument="--limit" type="integer" value="3" label="How many candidate regions to consider"/>\n-            <param argument="--species" type="text" optional="True" label="Name of existing Augustus species gene finding metaparameters"/>\n+\n+            <conditional name="aug_prediction">\n+                <param name="augustus_mode" type="select" label="Augustus species model">\n+                    <option value="no" selected="true">Use the default species for selected lineage</option>\n+                    <option value="builtin">Use another predefined species model</option>\n+                    <option value="history">Use a custom species model</option>\n+                </param>\n+                <when value="no"/>\n+                <when value="history">\n+                    <param name="augustus_model" type="data" format="augustus" label="Augustus model"/>\n+                </when>\n+                <when value="builtin">\n+                    <param name="augustus_species" type="select" label="Augustus species model">\n+                        <!-- If you update this list, please also update it in maker and augustus tools (../maker/maker.xml and ../augustus/augustus.xml) -->\n+                        <option value="human">Homo sapiens</option>\n+                        <option value="'..b'ns1">maker2_c_elegans1</option>\n+                        <option value="maker2_dmel1">maker2_dmel1</option>\n+                        <option value="maker2_spomb1">maker2_spomb1</option>\n+                        <option value="parasteatoda">parasteatoda</option>\n+                        <option value="rice">rice</option>\n+                        <option value="schistosoma2">schistosoma2</option>\n+                        <option value="sulfolobus_solfataricus">sulfolobus_solfataricus</option>\n+                    </param>\n+                </when>\n+            </conditional>\n             <param argument="--long" type="boolean" checked="false" truevalue="--long" falsevalue="" label="Optimization mode Augustus self-training" help="Adds considerably to run time, but can improve results for some non-model organisms"/>\n         </section>\n     </inputs>\n@@ -53,7 +180,7 @@\n     <tests>\n         <test>\n             <param name="input" value="genome.fa"/>\n-            <param name="lineage" value="arthropoda_2.0"/>\n+            <param name="lineage_path" value="arthropoda"/>\n             <param name="mode" value="geno"/>\n             <output name="busco_sum" file="genome_results/short_summary" compare="diff" lines_diff="4"/>\n             <output name="busco_table" file="genome_results/full_table" compare="diff" lines_diff="4"/>\n@@ -61,7 +188,7 @@\n         </test>\n         <test>\n             <param name="input" value="proteome.fa"/>\n-            <param name="lineage" value="arthropoda_2.0"/>\n+            <param name="lineage_path" value="arthropoda"/>\n             <param name="mode" value="prot"/>\n             <output name="busco_sum" file="proteome_results/short_summary" compare="diff" lines_diff="4"/>\n             <output name="busco_table" file="proteome_results/full_table" compare="diff" lines_diff="4"/>\n@@ -69,12 +196,32 @@\n         </test>\n         <test>\n             <param name="input" value="transcriptome.fa"/>\n-            <param name="lineage" value="arthropoda_2.0"/>\n+            <param name="lineage_path" value="arthropoda"/>\n             <param name="mode" value="tran"/>\n             <output name="busco_sum" file="transcriptome_results/short_summary" compare="diff" lines_diff="4"/>\n             <output name="busco_table" file="transcriptome_results/full_table" compare="diff" lines_diff="4"/>\n             <output name="busco_missing" file="transcriptome_results/missing_buscos_list" compare="diff" lines_diff="4"/>\n         </test>\n+        <test>\n+            <param name="input" value="genome.fa"/>\n+            <param name="lineage_path" value="arthropoda"/>\n+            <param name="mode" value="geno"/>\n+            <param name="adv|aug_prediction|augustus_mode" value="builtin"/>\n+            <param name="adv|aug_prediction|augustus_species" value="human"/>\n+            <output name="busco_sum" file="genome_results/short_summary" compare="diff" lines_diff="4"/>\n+            <output name="busco_table" file="genome_results/full_table" compare="diff" lines_diff="4"/>\n+            <output name="busco_missing" file="genome_results/missing_buscos_list" compare="diff" lines_diff="4"/>\n+        </test>\n+        <test>\n+            <param name="input" value="genome.fa"/>\n+            <param name="lineage_path" value="arthropoda"/>\n+            <param name="mode" value="geno"/>\n+            <param name="adv|aug_prediction|augustus_mode" value="history"/>\n+            <param name="adv|aug_prediction|augustus_model" value="local.tar.gz" ftype="augustus"/>\n+            <output name="busco_sum" file="genome_results/short_summary" compare="diff" lines_diff="4"/>\n+            <output name="busco_table" file="genome_results/full_table" compare="diff" lines_diff="4"/>\n+            <output name="busco_missing" file="genome_results/missing_buscos_list" compare="diff" lines_diff="4"/>\n+        </test>\n     </tests>\n     <help>\n         BUSCO: assessing genome assembly and annotation completeness with Benchmarking Universal Single-Copy Orthologs\n'
b
diff -r 87983967544b -r 382a4c19007f test-data/busco.loc
--- a/test-data/busco.loc Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/busco.loc Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,1 +1,1 @@
-arthropoda_2.0 arthropoda_2.0 Arthropoda (v2) ${__HERE__}/arthropoda/
+arthropoda Arthropoda data ${__HERE__}/
b
diff -r 87983967544b -r 382a4c19007f test-data/genome_results/full_table
--- a/test-data/genome_results/full_table Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/genome_results/full_table Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,5 +1,5 @@
-# BUSCO version is: 2.0 
-# The lineage dataset is: arthropoda (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
+# BUSCO version is: 3.0.2 
+# The lineage dataset is: N/A (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
 # To reproduce this run: python /tmp/tmpeQaU9k/job_working_directory/000/2/conda-env/bin/BUSCO.py -i /tmp/tmpeQaU9k/files/000/dataset_1.dat -o busco_galaxy -l /home/abretaud/iuc_tools_abretaud/tools/busco/test-data/arthropoda/ -m genome -c 1 -sp generic -e 0.01 -z
 #
 # Busco id Status Contig Start End Score Length
b
diff -r 87983967544b -r 382a4c19007f test-data/genome_results/missing_buscos_list
--- a/test-data/genome_results/missing_buscos_list Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/genome_results/missing_buscos_list Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,5 +1,5 @@
-# BUSCO version is: 2.0 
-# The lineage dataset is: arthropoda (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
+# BUSCO version is: 3.0.2 
+# The lineage dataset is: N/A (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
 # To reproduce this run: python /tmp/tmpeQaU9k/job_working_directory/000/2/conda-env/bin/BUSCO.py -i /tmp/tmpeQaU9k/files/000/dataset_1.dat -o busco_galaxy -l /home/abretaud/iuc_tools_abretaud/tools/busco/test-data/arthropoda/ -m genome -c 1 -sp generic -e 0.01 -z
 #
 BUSCOaEOG7B0HST
b
diff -r 87983967544b -r 382a4c19007f test-data/genome_results/short_summary
--- a/test-data/genome_results/short_summary Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/genome_results/short_summary Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,5 +1,5 @@
-# BUSCO version is: 2.0 
-# The lineage dataset is: arthropoda (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
+# BUSCO version is: 3.0.2 
+# The lineage dataset is: N/A (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
 # To reproduce this run: python /tmp/tmpeQaU9k/job_working_directory/000/2/conda-env/bin/BUSCO.py -i /tmp/tmpeQaU9k/files/000/dataset_1.dat -o busco_galaxy -l /home/abretaud/iuc_tools_abretaud/tools/busco/test-data/arthropoda/ -m genome -c 1 -sp generic -e 0.01 -z
 #
 # Summarized benchmarking in BUSCO notation for file /tmp/tmpeQaU9k/files/000/dataset_1.dat
b
diff -r 87983967544b -r 382a4c19007f test-data/local.tar.gz
b
Binary file test-data/local.tar.gz has changed
b
diff -r 87983967544b -r 382a4c19007f test-data/proteome_results/full_table
--- a/test-data/proteome_results/full_table Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/proteome_results/full_table Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,5 +1,5 @@
-# BUSCO version is: 2.0 
-# The lineage dataset is: arthropoda (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
+# BUSCO version is: 3.0.2 
+# The lineage dataset is: N/A (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
 # To reproduce this run: python /tmp/tmpCgx5rU/job_working_directory/000/4/conda-env/bin/BUSCO.py -i /tmp/tmpCgx5rU/files/000/dataset_5.dat -o busco_galaxy -l /home/abretaud/iuc_tools_abretaud/tools/busco/test-data/arthropoda/ -m proteins -c 1 -sp generic -e 0.01 -z
 #
 # Busco id Status Sequence Score Length
b
diff -r 87983967544b -r 382a4c19007f test-data/proteome_results/missing_buscos_list
--- a/test-data/proteome_results/missing_buscos_list Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/proteome_results/missing_buscos_list Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,4 +1,4 @@
-# BUSCO version is: 2.0 
-# The lineage dataset is: arthropoda (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
+# BUSCO version is: 3.0.2 
+# The lineage dataset is: N/A (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
 # To reproduce this run: python /tmp/tmpCgx5rU/job_working_directory/000/4/conda-env/bin/BUSCO.py -i /tmp/tmpCgx5rU/files/000/dataset_5.dat -o busco_galaxy -l /home/abretaud/iuc_tools_abretaud/tools/busco/test-data/arthropoda/ -m proteins -c 1 -sp generic -e 0.01 -z
 #
b
diff -r 87983967544b -r 382a4c19007f test-data/proteome_results/short_summary
--- a/test-data/proteome_results/short_summary Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/proteome_results/short_summary Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,5 +1,5 @@
-# BUSCO version is: 2.0 
-# The lineage dataset is: arthropoda (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
+# BUSCO version is: 3.0.2 
+# The lineage dataset is: N/A (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
 # To reproduce this run: python /tmp/tmpCgx5rU/job_working_directory/000/4/conda-env/bin/BUSCO.py -i /tmp/tmpCgx5rU/files/000/dataset_5.dat -o busco_galaxy -l /home/abretaud/iuc_tools_abretaud/tools/busco/test-data/arthropoda/ -m proteins -c 1 -sp generic -e 0.01 -z
 #
 # Summarized benchmarking in BUSCO notation for file /tmp/tmpCgx5rU/files/000/dataset_5.dat
b
diff -r 87983967544b -r 382a4c19007f test-data/transcriptome_results/full_table
--- a/test-data/transcriptome_results/full_table Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/transcriptome_results/full_table Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,5 +1,5 @@
-# BUSCO version is: 2.0 
-# The lineage dataset is: arthropoda (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
+# BUSCO version is: 3.0.2 
+# The lineage dataset is: N/A (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
 # To reproduce this run: python /tmp/tmpCgx5rU/job_working_directory/000/6/conda-env/bin/BUSCO.py -i /tmp/tmpCgx5rU/files/000/dataset_9.dat -o busco_galaxy -l /home/abretaud/iuc_tools_abretaud/tools/busco/test-data/arthropoda/ -m tran -c 1 -sp generic -e 0.01 -z
 #
 # Busco id Status Sequence Score Length
b
diff -r 87983967544b -r 382a4c19007f test-data/transcriptome_results/missing_buscos_list
--- a/test-data/transcriptome_results/missing_buscos_list Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/transcriptome_results/missing_buscos_list Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,5 +1,5 @@
-# BUSCO version is: 2.0 
-# The lineage dataset is: arthropoda (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
+# BUSCO version is: 3.0.2 
+# The lineage dataset is: N/A (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
 # To reproduce this run: python /tmp/tmpCgx5rU/job_working_directory/000/6/conda-env/bin/BUSCO.py -i /tmp/tmpCgx5rU/files/000/dataset_9.dat -o busco_galaxy -l /home/abretaud/iuc_tools_abretaud/tools/busco/test-data/arthropoda/ -m tran -c 1 -sp generic -e 0.01 -z
 #
 BUSCOaEOG7B0HST
b
diff -r 87983967544b -r 382a4c19007f test-data/transcriptome_results/short_summary
--- a/test-data/transcriptome_results/short_summary Tue Jan 24 12:08:38 2017 -0500
+++ b/test-data/transcriptome_results/short_summary Fri Oct 20 03:54:35 2017 -0400
[
@@ -1,9 +1,9 @@
-# BUSCO version is: 2.0 
-# The lineage dataset is: arthropoda (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
+# BUSCO version is: 3.0.2 
+# The lineage dataset is: N/A (Creation date: N/A, number of species: N/A, number of BUSCOs: N/A)
 # To reproduce this run: python /tmp/tmpCgx5rU/job_working_directory/000/6/conda-env/bin/BUSCO.py -i /tmp/tmpCgx5rU/files/000/dataset_9.dat -o busco_galaxy -l /home/abretaud/iuc_tools_abretaud/tools/busco/test-data/arthropoda/ -m tran -c 1 -sp generic -e 0.01 -z
 #
 # Summarized benchmarking in BUSCO notation for file /tmp/tmpCgx5rU/files/000/dataset_9.dat
-# BUSCO was run in mode: tran
+# BUSCO was run in mode: transcriptome
 
  C:0.0%[S:0.0%,D:0.0%],F:0.0%,M:100.0%,n:1
 
b
diff -r 87983967544b -r 382a4c19007f tool-data/busco.loc.sample
--- a/tool-data/busco.loc.sample Tue Jan 24 12:08:38 2017 -0500
+++ b/tool-data/busco.loc.sample Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,13 +1,13 @@
 # This is a sample file distributed with Galaxy that is used to define a
 # list of busco datasets, using four columns tab separated:
 #
-# <unique_build_id> <dbkey> <display_name> <genome_fasta_file_path>
+# <unique_build_id> <display_name> <genome_fasta_file_path>
 #
 # Datasets can be retrieved from http://busco.ezlab.org/frame_wget.html
 #
 # "/some/path/arthropoda/" would be the last column in the line
 # If this were for the mm10 mouse genome, the resulting entry would look like:
 #
-#arthropoda_2.0 arthropoda_2.0 Arthropoda (v2) /some/path/arthropoda/
+#arthropoda_2.0 arthropoda_2.0 /some/path/arthropoda/
 #
 #
b
diff -r 87983967544b -r 382a4c19007f tool_data_table_conf.xml.sample
--- a/tool_data_table_conf.xml.sample Tue Jan 24 12:08:38 2017 -0500
+++ b/tool_data_table_conf.xml.sample Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,6 +1,6 @@
 <tables>
     <table name="busco" comment_char="#">
-        <columns>value, dbkey, name, path</columns>
+        <columns>value, name, path</columns>
         <file path="tool-data/busco.loc" />
     </table>
 </tables>
b
diff -r 87983967544b -r 382a4c19007f tool_data_table_conf.xml.test
--- a/tool_data_table_conf.xml.test Tue Jan 24 12:08:38 2017 -0500
+++ b/tool_data_table_conf.xml.test Fri Oct 20 03:54:35 2017 -0400
b
@@ -1,6 +1,6 @@
 <tables>
     <table name="busco" comment_char="#">
-        <columns>value, dbkey, name, path</columns>
+        <columns>value, name, path</columns>
         <file path="${__HERE__}/test-data/busco.loc" />
     </table>
 </tables>