update BUSCO and fix --miniprot parameter (#6153)

* update BUSCO and fiw --miniprot parameter * error version * update * fix miniprot parameter * update test-data * update * small modification on busco db * small modifications * fix output errors * small modification * small modification * fix test * add test-data/genome_results_miniprot * small change in busco.xml * update test-data * add assert * fix test 7
galaxyproject · Aug 30, 2024 · 21578f9 · 21578f9
1 parent 3d970de
commit 21578f9
Show file tree

Hide file tree

Showing 32 changed files with 1,452 additions and 180 deletions.
diff --git a/tools/busco/busco.xml b/tools/busco/busco.xml
diff --git a/tools/busco/macros.xml b/tools/busco/macros.xml
@@ -1,6 +1,6 @@
 <?xml version="1.0"?>
 <macros>
-    <token name="@TOOL_VERSION@">5.5.0</token>
+    <token name="@TOOL_VERSION@">5.7.1</token>
     <token name="@VERSION_SUFFIX@">0</token>
 
     <xml name="citations">

diff --git a/tools/busco/test-data/busco_database.loc b/tools/busco/test-data/busco_database.loc
@@ -3,4 +3,4 @@
 # - name
 # - version
 # - /path/to/data 
-busco-demo-db-20230328	BUSCO-DEMO-DB-20230328	5.4.6	${__HERE__}/test-db/busco_downloads
+busco-demo-db-20230328	BUSCO-DEMO-DB-20230328	5.4.6	${__HERE__}/test-db/busco_downloads
diff --git a/tools/busco/test-data/busco_downloads/file_versions.tsv b/tools/busco/test-data/busco_downloads/file_versions.tsv
diff --git a/tools/busco/test-data/genome_results/full_table b/tools/busco/test-data/genome_results/full_table
@@ -1,5 +1,5 @@
-# BUSCO version is: 5.5.0 
-# The lineage dataset is: arthropoda_odb10 (Creation date: 2020-09-10, number of genomes: 90, number of BUSCOs: 1013)
+# BUSCO version is: 5.7.1 
+# The lineage dataset is: arthropoda_odb10 (Creation date: 2024-01-08, number of genomes: 90, number of BUSCOs: 1013)
 # Busco id	Status	Sequence	Gene Start	Gene End	Strand	Score	Length	OrthoDB url	Description
 774at6656	Missing
 980at6656	Missing
@@ -509,7 +509,7 @@
 93535at6656	Missing
 93797at6656	Missing
 94054at6656	Missing
-94238at6656	Complete	sample:34764-38486	34764	38486	-	60.7	116	https://www.orthodb.org/v10?query=94238at6656	checkpoint protein HUS1
+94238at6656	Complete	sample	38486	34764	-	60.7	116	https://v10-1.orthodb.org/?query=94238at6656	Checkpoint protein HUS1
 94263at6656	Missing
 94304at6656	Missing
 94473at6656	Missing

diff --git a/tools/busco/test-data/genome_results/full_table_cached b/tools/busco/test-data/genome_results/full_table_cached
@@ -1,4 +1,4 @@
-# BUSCO version is: 5.5.0 
+# BUSCO version is: 5.7.1 
 # The lineage dataset is: archaea_odb10 (Creation date: 2021-02-23, number of genomes: 404, number of BUSCOs: 194)
 # Busco id	Status	Sequence	Gene Start	Gene End	Strand	Score	Length
 860at2157	Missing

diff --git a/tools/busco/test-data/genome_results/missing_buscos_list b/tools/busco/test-data/genome_results/missing_buscos_list
@@ -1,5 +1,5 @@
-# BUSCO version is: 5.5.0 
-# The lineage dataset is: arthropoda_odb10 (Creation date: 2020-09-10, number of genomes: 90, number of BUSCOs: 1013)
+# BUSCO version is: 5.7.1 
+# The lineage dataset is: arthropoda_odb10 (Creation date: 2024-01-08, number of genomes: 90, number of BUSCOs: 1013)
 # Busco id
 100070at6656
 100136at6656

diff --git a/tools/busco/test-data/genome_results/short_summary b/tools/busco/test-data/genome_results/short_summary
@@ -1,35 +1,36 @@
-# BUSCO version is: 5\.5\.0 
-# The lineage dataset is: arthropoda_odb10 \(Creation date: [0-9]{4}-[0-9]{2}-[0-9]{2}, number of genomes: 90, number of BUSCOs: 1013\)
-# Summarized benchmarking in BUSCO notation for file [a-z0-9_\-/\.]+
+# BUSCO version is: 5.7.1 
+# The lineage dataset is: arthropoda_odb10 (Creation date: 2024-01-08, number of genomes: 90, number of BUSCOs: 1013)
+# Summarized benchmarking in BUSCO notation for file /tmp/tmpl5l1blpe/files/7/a/3/dataset_7a33f452-1064-4b4a-943f-b0efef6a4a4a.dat
 # BUSCO was run in mode: euk_genome_aug
 # Gene predictor used: augustus
 
-	\*\*\*\*\* Results: \*\*\*\*\*
+	***** Results: *****
 
-	C:0\.1%\[S:0\.1%,D:0\.0%\],F:0\.0%,M:99\.9%,n:1013	   
-	1	Complete BUSCOs \(C\)			   
-	1	Complete and single-copy BUSCOs \(S\)	   
-	0	Complete and duplicated BUSCOs \(D\)	   
-	0	Fragmented BUSCOs \(F\)			   
-	1012	Missing BUSCOs \(M\)			   
+	C:0.1%[S:0.1%,D:0.0%],F:0.0%,M:99.9%,n:1013	   
+	1	Complete BUSCOs (C)			   
+	1	Complete and single-copy BUSCOs (S)	   
+	0	Complete and duplicated BUSCOs (D)	   
+	0	Fragmented BUSCOs (F)			   
+	1012	Missing BUSCOs (M)			   
 	1013	Total BUSCO groups searched		   
 
 Assembly Statistics:
 	1	Number of scaffolds
 	1	Number of contigs
 	62370	Total length
-	0\.000%	Percent gaps
+	0.000%	Percent gaps
 	62 KB	Scaffold N50
 	62 KB	Contigs N50
 
 
 Dependencies and versions:
-	hmmsearch: [0-9\.\+]+
-	bbtools: [0-9\.\+]+
-	makeblastdb: [0-9\.\+]+
-	tblastn: [0-9\.\+]+
-	augustus: [0-9\.\+]+
-	gff2gbSmallDNA\.pl: None
-	new_species\.pl: None
+	hmmsearch: 3.1
+	bbtools: 39.06
+	makeblastdb: 2.15.0+
+	tblastn: 2.15.0+
+	augustus: 3.5.0
+	gff2gbSmallDNA.pl: None
+	new_species.pl: None
 	etraining: None
-	busco: [0-9\.\+]+
+	python: sys.version_info(major=3, minor=9, micro=19, releaselevel='final', serial=0)
+	busco: 5.7.1
diff --git a/tools/busco/test-data/genome_results/short_summary_cached b/tools/busco/test-data/genome_results/short_summary_cached
@@ -1,7 +1,7 @@
-# BUSCO version is: 5.5.0 
+# BUSCO version is: 5.7.1 
 # The lineage dataset is: archaea_odb10 (Creation date: 2021-02-23, number of genomes: 404, number of BUSCOs: 194)
-# Summarized benchmarking in BUSCO notation for file /tmp/tmpbmq1q2c6/files/3/3/0/dataset_330f49ed-7a6b-4380-870f-22494bb4b257.dat
-# BUSCO was run in mode: prok_genome
+# Summarized benchmarking in BUSCO notation for file /tmp/tmprlx1s52v/files/e/8/d/dataset_e8d136af-46c1-4ffa-abaf-e8899526d2b0.dat
+# BUSCO was run in mode: prok_genome_prod
 # Gene predictor used: prodigal
 
 	***** Results: *****
@@ -27,4 +27,5 @@ Dependencies and versions:
 	hmmsearch: 3.1
 	bbtools: 39.01
 	prodigal: 2.6.3
-	busco: 5.5.0
+	python: sys.version_info(major=3, minor=9, micro=19, releaselevel='final', serial=0)
+	busco: 5.7.1
diff --git a/tools/busco/test-data/genome_results_metaeuk/full_table b/tools/busco/test-data/genome_results_metaeuk/full_table
@@ -1,5 +1,5 @@
-# BUSCO version is: 5.5.0 
-# The lineage dataset is: arthropoda_odb10 (Creation date: 2020-09-10, number of genomes: 90, number of BUSCOs: 1013)
+# BUSCO version is: 5.7.1 
+# The lineage dataset is: arthropoda_odb10 (Creation date: 2024-01-08, number of genomes: 90, number of BUSCOs: 1013)
 # Busco id	Status	Sequence	Gene Start	Gene End	Strand	Score	Length	OrthoDB url	Description
 774at6656	Missing
 980at6656	Missing
@@ -340,7 +340,7 @@
 68939at6656	Missing
 68961at6656	Missing
 68981at6656	Missing
-68987at6656	Complete	sample:40255-42070	40255	42070	+	122.8	266	https://www.orthodb.org/v10?query=68987at6656	mannose-1-phosphate guanyltransferase alpha
+68987at6656	Complete	sample	40255	42070	+	122.8	266	https://v10-1.orthodb.org/?query=68987at6656	Nucleotidyl transferase domain
 69201at6656	Missing
 69238at6656	Missing
 69284at6656	Missing
@@ -509,7 +509,7 @@
 93535at6656	Missing
 93797at6656	Missing
 94054at6656	Missing
-94238at6656	Complete	sample:35678-34845	34845	35678	-	60.7	116	https://www.orthodb.org/v10?query=94238at6656	checkpoint protein HUS1
+94238at6656	Complete	sample	35693	34845	-	60.6	116	https://v10-1.orthodb.org/?query=94238at6656	Checkpoint protein HUS1
 94263at6656	Missing
 94304at6656	Missing
 94473at6656	Missing

diff --git a/tools/busco/test-data/genome_results_metaeuk/missing_buscos_list b/tools/busco/test-data/genome_results_metaeuk/missing_buscos_list
@@ -1,5 +1,5 @@
-# BUSCO version is: 5.5.0 
-# The lineage dataset is: arthropoda_odb10 (Creation date: 2020-09-10, number of genomes: 90, number of BUSCOs: 1013)
+# BUSCO version is: 5.7.1 
+# The lineage dataset is: arthropoda_odb10 (Creation date: 2024-01-08, number of genomes: 90, number of BUSCOs: 1013)
 # Busco id
 100070at6656
 100136at6656

diff --git a/tools/busco/test-data/genome_results_metaeuk/out.gff3 b/tools/busco/test-data/genome_results_metaeuk/out.gff3
@@ -0,0 +1,3 @@
+##gff-version 3
+sample	MetaEuk	gene	40256	42071	198	+	.	Target_ID=68987at6656_29053_0:00071c;TCS_ID=68987at6656_29053_0:00071c|sample|+|40255
+sample	MetaEuk	gene	34846	35694	527	-	.	Target_ID=94238at6656_7245_0:00200b;TCS_ID=94238at6656_7245_0:00200b|sample|-|34845
diff --git a/tools/busco/test-data/genome_results_metaeuk/short_summary b/tools/busco/test-data/genome_results_metaeuk/short_summary
@@ -1,30 +1,32 @@
-# BUSCO version is: 5\.5\.0 
-# The lineage dataset is: arthropoda_odb10 \(Creation date: [0-9]{4}-[0-9]{2}-[0-9]{2}, number of genomes: 90, number of BUSCOs: 1013\)
-# Summarized benchmarking in BUSCO notation for file [a-z0-9_\-/\.]+
-# BUSCO was run in mode: euk_genome_met
-# Gene predictor used: metaeuk
+# BUSCO version is: 5.7.1 
+# The lineage dataset is: arthropoda_odb10 (Creation date: 2024-01-08, number of genomes: 90, number of BUSCOs: 1013)
+# Summarized benchmarking in BUSCO notation for file /tmp/tmpl5l1blpe/files/f/3/1/dataset_f31d44e3-c824-4cdf-92ba-99a2c26071d2.dat
+# BUSCO was run in mode: euk_genome_min
+# Gene predictor used: miniprot
 
-	\*\*\*\*\* Results: \*\*\*\*\*
+	***** Results: *****
 
-	C:0\.2%\[S:0\.2%,D:0\.0%\],F:0\.0%,M:99\.8%,n:1013	   
-	2	Complete BUSCOs \(C\)			   
-	2	Complete and single-copy BUSCOs \(S\)	   
-	0	Complete and duplicated BUSCOs \(D\)	   
-	0	Fragmented BUSCOs \(F\)			   
-	1011	Missing BUSCOs \(M\)			   
+	C:0.1%[S:0.1%,D:0.0%],F:0.0%,M:99.9%,n:1013	   
+	1	Complete BUSCOs (C)			   
+	1	Complete and single-copy BUSCOs (S)	   
+	0	Complete and duplicated BUSCOs (D)	   
+	0	Fragmented BUSCOs (F)			   
+	1012	Missing BUSCOs (M)			   
 	1013	Total BUSCO groups searched		   
 
 Assembly Statistics:
 	1	Number of scaffolds
 	1	Number of contigs
 	62370	Total length
-	0\.000%	Percent gaps
+	0.000%	Percent gaps
 	62 KB	Scaffold N50
 	62 KB	Contigs N50
 
 
 Dependencies and versions:
-	hmmsearch: [0-9\.\+]+
-	bbtools: [0-9\.\+]+
-	metaeuk: [0-9a-z\.\+]+
-	busco: [0-9\.\+]+
+	hmmsearch: 3.1
+	bbtools: 39.06
+	miniprot_index: 0.13-r248
+	miniprot_align: 0.13-r248
+	python: sys.version_info(major=3, minor=9, micro=19, releaselevel='final', serial=0)
+	busco: 5.7.1
diff --git a/tools/busco/test-data/genome_results_metaeuk/summary.png b/tools/busco/test-data/genome_results_metaeuk/summary.png
diff --git a/tools/busco/test-data/genome_results_metaeuk_auto/full_table b/tools/busco/test-data/genome_results_metaeuk_auto/full_table
@@ -1,6 +1,6 @@
-# BUSCO version is: 5.5.0 
-# The lineage dataset is: eukaryota_odb10 (Creation date: 2020-09-10, number of genomes: 70, number of BUSCOs: 255)
-# Busco id	Status	Sequence	Gene Start	Gene End	Strand	Score	Length
+# BUSCO version is: 5.7.1 
+# The lineage dataset is: eukaryota_odb10 (Creation date: 2024-01-08, number of genomes: 70, number of BUSCOs: 255)
+# Busco id	Status	Sequence	Gene Start	Gene End	Strand	Score	Length	OrthoDB url	Description
 39650at2759	Missing
 83779at2759	Missing
 87842at2759	Missing
@@ -131,7 +131,7 @@
 1041560at2759	Missing
 1049599at2759	Missing
 1051021at2759	Missing
-1053181at2759	Complete	sample:35678-34845	34845	35678	-	45.0	149
+1053181at2759	Complete	sample	35693	34845	-	44.7	149	https://v10-1.orthodb.org/?query=1053181at2759	Checkpoint protein Hus1/Mec3
 1057950at2759	Missing
 1065019at2759	Missing
 1076134at2759	Missing

diff --git a/tools/busco/test-data/genome_results_metaeuk_auto/missing_buscos_list b/tools/busco/test-data/genome_results_metaeuk_auto/missing_buscos_list
@@ -1,5 +1,5 @@
-# BUSCO version is: 5.5.0 
-# The lineage dataset is: eukaryota_odb10 (Creation date: 2020-09-10, number of genomes: 70, number of BUSCOs: 255)
+# BUSCO version is: 5.7.1 
+# The lineage dataset is: eukaryota_odb10 (Creation date: 2024-01-08, number of genomes: 70, number of BUSCOs: 255)
 # Busco id
 1001705at2759
 1003258at2759

diff --git a/tools/busco/test-data/genome_results_metaeuk_auto/out.gff b/tools/busco/test-data/genome_results_metaeuk_auto/out.gff
@@ -1,5 +1,2 @@
 ##gff-version 3
 sample	MetaEuk	gene	34846	35694	527	-	.	Target_ID=1053181at2759_7245_0:00200b;TCS_ID=1053181at2759_7245_0:00200b|sample|-|34845
-sample	MetaEuk	mRNA	34846	35694	527	-	.	Target_ID=1053181at2759_7245_0:00200b;TCS_ID=1053181at2759_7245_0:00200b|sample|-|34845_mRNA;Parent=1053181at2759_7245_0:00200b|sample|-|34845
-sample	MetaEuk	exon	34846	35694	527	-	.	Target_ID=1053181at2759_7245_0:00200b;TCS_ID=1053181at2759_7245_0:00200b|sample|-|34845_exon_0;Parent=1053181at2759_7245_0:00200b|sample|-|34845_mRNA
-sample	MetaEuk	CDS	34846	35694	527	-	.	Target_ID=1053181at2759_7245_0:00200b;TCS_ID=1053181at2759_7245_0:00200b|sample|-|34845_CDS_0;Parent=1053181at2759_7245_0:00200b|sample|-|34845_exon_0
diff --git a/tools/busco/test-data/genome_results_metaeuk_auto/short_summary b/tools/busco/test-data/genome_results_metaeuk_auto/short_summary
@@ -1,6 +1,6 @@
-# BUSCO version is: 5.5.0 
-# The lineage dataset is: eukaryota_odb10 (Creation date: 2020-09-10, number of genomes: 70, number of BUSCOs: 255)
-# Summarized benchmarking in BUSCO notation for file /tmp/tmp5_syjrgy/files/a/5/a/dataset_a5adf25d-f667-41d6-9472-f8661403e128.dat
+# BUSCO version is: 5.7.1 
+# The lineage dataset is: eukaryota_odb10 (Creation date: 2024-01-08, number of genomes: 70, number of BUSCOs: 255)
+# Summarized benchmarking in BUSCO notation for file /tmp/tmpg41og118/files/f/5/e/dataset_f5e13834-6f57-4cb4-af82-b730a3e03fdb.dat
 # BUSCO was run in mode: euk_genome_met
 # Gene predictor used: metaeuk
 
@@ -25,7 +25,8 @@ Assembly Statistics:
 
 Dependencies and versions:
 	hmmsearch: 3.1
-	bbtools: 39.01
+	bbtools: 39.06
 	prodigal: 2.6.3
-	busco: 5.5.0
-	metaeuk: 6.a5d39d9
+	python: sys.version_info(major=3, minor=9, micro=19, releaselevel='final', serial=0)
+	busco: 5.7.1
+	metaeuk: 7.bba0d80
diff --git a/tools/busco/test-data/genome_results_metaeuk_auto/summary.png b/tools/busco/test-data/genome_results_metaeuk_auto/summary.png