Updated uniprot

Updated interpro
IGBIllinois · Sep 9, 2024 · 34d46c8 · 34d46c8
1 parent da184c9
commit 34d46c8
Show file tree

Hide file tree

Showing 19 changed files with 316 additions and 7 deletions.
diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
diff --git a/.gitignore b/.gitignore
diff --git a/LICENSE b/LICENSE
diff --git a/README.md b/README.md
diff --git a/easyconfigs/b/BUSCO-db/BUSCO-db_extract.sh b/easyconfigs/b/BUSCO-db/BUSCO-db_extract.sh
@@ -0,0 +1,58 @@
+#!/bin/bash
+# ----------------SLURM Parameters----------------
+#SBATCH -p admin
+#SBATCH -n 4
+#SBATCH --mem=20g
+#SBATCH -N 1
+#SBATCH --mail-user=datamover@igb.illinois.edu
+#SBATCH --mail-type=ALL
+#SBATCH -J BUSCO-db_extract
+#SBATCH -D /home/a-m/datamover/jobs
+#SBATCH -o %x-%j.out
+# ----------------Load Modules--------------------
+module load pigz/2.4-IGB-gcc-8.2.0
+# ----------------Commands------------------------
+#
+# Replace DATABASE with name of database you are downloading
+# Replace WEBSITE with remote location of database#
+#
+
+DATABASE="BUSCO-db"
+
+if [ -z "$1" ];
+then
+	echo "Please specify ${DATABASE} version number";
+	exit 1;
+fi
+
+VERSION=$1
+MIRROR_DIR=/private_stores/mirror/${DATABASE}/${VERSION}
+
+
+echo "`date "+%Y-%m-%d %k:%M:%S"` Extracting tar.gz Files"
+for f in $(find ${MIRROR_DIR} -name '*.gz');
+do
+	gzip -d $f
+	if [ $? -ne 0 ]; then
+                echo "`date "+%Y-%m-%d %k:%M:%S"` Error extracting file: $f"
+                exit $?
+        else
+                echo "`date "+%Y-%m-%d %k:%M:%S"` Done extracting file: $f"
+        fi
+done
+
+echo "`date "+%Y-%m-%d %k:%M:%S"` Extracting .gz Files with Pigz"
+pigz -p ${SLURM_NTASKS} -dr ${MIRROR_DIR}
+if [ $? -ne 0 ]
+then
+	echo "`date "+%Y-%m-%d %k:%M:%S"` Extracting files Failed"
+	exit $?
+else
+	echo "`date "+%Y-%m-%d %k:%M:%S"` Extracting Files Complete"
+fi
+
+echo "`date "+%Y-%m-%d %k:%M:%S"` Fix Permissions Start"
+find ${MIRROR_DIR} -type d -exec chmod 775 {} \;
+find ${MIRROR_DIR} -type f -exec chmod 664 {} \;
+echo "`date "+%Y-%m-%d %k:%M:%S"` Fix Permissions Completed"
+
diff --git a/easyconfigs/c/checkv-db/checkv-db-1.5.eb b/easyconfigs/c/checkv-db/checkv-db-1.5.eb
@@ -0,0 +1,27 @@
+easyblock = 'Bundle'
+
+name = 'checkv-db'
+version = '1.5'
+
+homepage = 'https://portal.nersc.gov/CheckV/'
+description = """The main change in v1.5 was to remove putatively non-viral sequences to minimize high-confidence matches to the CheckV database for non-viral sequences."""
+
+toolchain = SYSTEM
+
+
+dependencies = []
+
+sanity_check_paths = {
+        'files': [],
+        'dirs': ['checkv-db-v1.5'],
+}
+
+modextrapaths = {
+	'BIODB':''
+}
+
+keeppreviousinstall = True
+moduleclass = 'data'
+
+modloadmsg = "%(name)s/%(version)s database is located at %(installdir)s\n"
+
diff --git a/easyconfigs/c/checkv-db/checkv-db_download.sh b/easyconfigs/c/checkv-db/checkv-db_download.sh
@@ -0,0 +1,42 @@
+#!/bin/bash
+# ----------------SLURM Parameters----------------
+#SBATCH -p admin
+#SBATCH -n 1
+#SBATCH --mem=20g
+#SBATCH -N 1
+#SBATCH --mail-user=datamover@igb.illinois.edu
+#SBATCH --mail-type=ALL
+#SBATCH -J checkv-db_download
+#SBATCH -D /home/a-m/datamover/jobs
+#SBATCH -o %x-%j.out
+# ----------------Load Modules--------------------
+# ----------------Commands------------------------
+#
+# Replace DATABASE with name of database you are downloading
+# Replace WEBSITE with remote location of database
+#
+
+DATABASE="checkv-db"
+
+if [ -z "$1" ];
+then
+	echo "Please specify ${DATABASE} version number";
+	exit 1;
+fi
+
+VERSION=$1
+MIRROR_DIR=/private_stores/mirror/${DATABASE}/${VERSION}
+
+echo "`date "+%Y-%m-%d %k:%M:%S"` Downloading Files"
+mkdir -p ${MIRROR_DIR}
+wget https://portal.nersc.gov/CheckV/checkv-db-v${VERSION}.tar.gz -P ${MIRROR_DIR}/
+if [ $? -ne 0 ]
+then
+	echo "`date "+%Y-%m-%d %k:%M:%S"` Downloading Files Failed"
+	exit $?
+else
+	echo "`date "+%Y-%m-%d %k:%M:%S"` Downloading Files Complete"
+fi
+
+
+
diff --git a/easyconfigs/c/checkv-db/checkv-db_extract.sh b/easyconfigs/c/checkv-db/checkv-db_extract.sh
@@ -0,0 +1,50 @@
+#!/bin/bash
+# ----------------SLURM Parameters----------------
+#SBATCH -p admin
+#SBATCH -n 4
+#SBATCH --mem=20g
+#SBATCH -N 1
+#SBATCH --mail-user=datamover@igb.illinois.edu
+#SBATCH --mail-type=ALL
+#SBATCH -J checkv-db_extract
+#SBATCH -D /home/a-m/datamover/jobs
+#SBATCH -o %x-%j.out
+# ----------------Load Modules--------------------
+module load pigz/2.4-IGB-gcc-8.2.0
+# ----------------Commands------------------------
+#
+# Replace DATABASE with name of database you are downloading
+# Replace WEBSITE with remote location of database#
+#
+
+DATABASE="checkv-db"
+
+if [ -z "$1" ];
+then
+	echo "Please specify ${DATABASE} version number";
+	exit 1;
+fi
+
+VERSION=$1
+MIRROR_DIR=/private_stores/mirror/${DATABASE}/${VERSION}
+
+
+echo "`date "+%Y-%m-%d %k:%M:%S"` Extracting tar.gz Files"
+for f in $(find ${MIRROR_DIR} -name '*.tar.gz');
+do
+	tar -xvzf $f -C `dirname $f`
+	if [ $? -ne 0 ]; then
+                echo "`date "+%Y-%m-%d %k:%M:%S"` Error extracting file: $f"
+                exit $?
+        else
+                echo "`date "+%Y-%m-%d %k:%M:%S"` Done extracting file: $f"
+        fi
+done
+
+#Run diamond to create index
+diamond makedb -p $SLURM_NTASKS --in ${MIRROR_DIR}/checkv-db-v${VERSION}/genome_db/checkv_reps.faa --db ${MIRROR_DIR}/checkv-db-v${VERSION}/genome_db/checkv_reps.dmnd
+echo "`date "+%Y-%m-%d %k:%M:%S"` Fix Permissions Start"
+find ${MIRROR_DIR} -type d -exec chmod 775 {} \;
+find ${MIRROR_DIR} -type f -exec chmod 664 {} \;
+echo "`date "+%Y-%m-%d %k:%M:%S"` Fix Permissions Completed"
+
diff --git a/easyconfigs/d/dorado-db/dorado-db-20240501.eb b/easyconfigs/d/dorado-db/dorado-db-20240501.eb
@@ -0,0 +1,29 @@
+easyblock = 'Bundle'
+
+name = 'dorado-db'
+version = '20240501'
+
+homepage = 'https://github.com/nanoporetech/dorado'
+description = """Databases for dorado"""
+
+toolchain = SYSTEM
+
+source_urls = ['']
+sources = []
+
+dependencies = []
+
+sanity_check_paths = {
+        'files': [],
+        'dirs': ['dna_r9.4.1_e8_fast@v3.4'],
+}
+
+modextrapaths = {
+	'BIODB':''
+}
+
+keeppreviousinstall = True
+moduleclass = 'data'
+
+modloadmsg = "%(name)s/%(version)s database is located at %(installdir)s\n"
+
diff --git a/easyconfigs/d/dorado-db/dorado-db_download.sh b/easyconfigs/d/dorado-db/dorado-db_download.sh
@@ -0,0 +1,35 @@
+#!/bin/bash
+# ----------------SLURM Parameters----------------
+#SBATCH -p admin
+#SBATCH -n 1
+#SBATCH --mem=20g
+#SBATCH -N 1
+#SBATCH --mail-user=datamover@igb.illinois.edu
+#SBATCH --mail-type=ALL
+#SBATCH -J dorado-db_download
+#SBATCH -D /home/a-m/datamover/jobs
+#SBATCH -o %x-%j.out
+# ----------------Load Modules--------------------
+module load dorado
+# ----------------Commands------------------------
+#
+# Replace DATABASE with name of database you are downloading
+# Replace WEBSITE with remote location of database
+#
+
+DATABASE="dorado-db"
+
+if [ -z "$1" ];
+then
+	echo "Please specify ${DATABASE} version number";
+	exit 1;
+fi
+
+VERSION=$1
+MIRROR_DIR=/private_stores/mirror/${DATABASE}/${VERSION}
+
+echo "`date "+%Y-%m-%d %k:%M:%S"` Downloading Files"
+mkdir -p ${MIRROR_DIR}
+cd  ${MIRROR_DIR}
+dorado download
+
diff --git a/easyconfigs/f/funannotate-db/funannotate-db-20240515.eb b/easyconfigs/f/funannotate-db/funannotate-db-20240515.eb
@@ -0,0 +1,30 @@
+easyblock = 'Bundle'
+
+name = 'funannotate-db'
+version = '20240515'
+
+homepage = 'https://github.com/nextgenusfs/funannotate'
+description = """funannotate is a pipeline for genome annotation (built specifically for fungi, but will also work with higher eukaryotes). Installation, usage, and more information can be found at http://funannotate.readthedocs.io"""
+
+toolchain = SYSTEM
+
+source_urls = ['']
+sources = []
+
+dependencies = []
+
+sanity_check_paths = {
+        'files': ['uniprot.dmnd'],
+        'dirs': [],
+}
+
+modextrapaths = {
+	'BIODB':'',
+	'FUNANNOTATE_DB': ''
+}
+
+keeppreviousinstall = True
+moduleclass = 'data'
+
+modloadmsg = "%(name)s/%(version)s database is located at %(installdir)s\n"
+
diff --git a/easyconfigs/i/interpro/interpro_download.sh b/easyconfigs/i/interpro/interpro_download.sh
@@ -1,5 +1,5 @@
 # ----------------Load Modules--------------------
-module load globus-cli/3.18.0-IGB-gcc-8.2.0-Python-3.7.2
+module load globus-cli/3.30.1-IGB-gcc-8.2.0-Python-3.10.1
 
 DATABASE="interpro"
 
@@ -14,6 +14,6 @@ MIRROR_DIR=/private_stores/mirror/${DATABASE}/${VERSION}
 
 echo "`date "+%Y-%m-%d %k:%M:%S"` Downloading Files"
 mkdir -p ${MIRROR_DIR}
-globus transfer -r --exclude "reference_proteomes" --exclude "taxonomic_divisions" "47772002-3e5b-4fd3-b97c-18cee38d6df2:/pub/databases/interpro/releases/$VERSION/" "1ccc563b-0542-44e5-a13c-fc4b00281b72:${MIRROR_DIR}/"
+globus transfer -r --exclude "reference_proteomes" --exclude "taxonomic_divisions" "47772002-3e5b-4fd3-b97c-18cee38d6df2:/pub/databases/interpro/releases/$VERSION/" "4a467fda-f559-4fc3-b54a-e2842f439e06:${MIRROR_DIR}/"
 
 
diff --git a/easyconfigs/p/pgap-db/pgap-db-2021-07-01.build5508 b/easyconfigs/p/pgap-db/pgap-db-2021-07-01.build5508
diff --git a/easyconfigs/r/R/R_mirror_update.pl b/easyconfigs/r/R/R_mirror_update.pl
@@ -7,7 +7,7 @@
 my $cran_dir="/private_stores/mirror/R/cran";
 my $cran_exclude="--exclude='bin'";
 
-my @bioc_versions = ('3.18');
+my @bioc_versions = ('3.19');
 
 foreach my $version (@bioc_versions) {
 

diff --git a/easyconfigs/templates/globus.sh b/easyconfigs/templates/globus.sh
@@ -13,7 +13,7 @@ VERSION=$1
 DATABASE=""
 MIRROR_DIR=/private_stores/mirror/${DATABASE}/${VERSION}
 FASTA_DIR=${MIRROR_DIR}/db
-BIOTRANSFER_UUID="1ccc563b-0542-44e5-a13c-fc4b00281b72"
+BIOTRANSFER_UUID="4a467fda-f559-4fc3-b54a-e2842f439e06"
 SOURCE_UUID=""
 
 echo "Downloading Files: `date "+%Y-%m-%d %k:%M:%S"`"

diff --git a/easyconfigs/u/uniprot/.uniprot_download.sh.swp b/easyconfigs/u/uniprot/.uniprot_download.sh.swp
diff --git a/easyconfigs/u/uniprot/uniprot-2023_05.eb b/easyconfigs/u/uniprot/uniprot-2023_05.eb
@@ -1,7 +1,7 @@
 easyblock = 'Bundle'
 
 name = 'uniprot'
-version = '2023_04'
+version = '2023_05'
 
 
 homepage = 'https://www.uniprot.org/'

diff --git a/easyconfigs/u/uniprot/uniprot_2024_04.eb b/easyconfigs/u/uniprot/uniprot_2024_04.eb
@@ -0,0 +1,38 @@
+easyblock = 'Bundle'
+
+name = 'uniprot'
+version = '2024_04'
+
+
+homepage = 'https://www.uniprot.org/'
+description = """The mission of UniProt is to provide the scientific community with a comprehensive, high-quality and freely accessible resource of protein sequence and functional information. """
+
+toolchain = SYSTEM
+
+dependencies = []
+
+
+sanity_check_paths = {
+        'files': ['db/knowledgebase/complete/uniprot_sprot.fasta',
+		'db/knowledgebase/complete/uniprot_trembl.fasta',
+		'db/knowledgebase/idmapping/idmapping.dat',
+		'db/uniref/uniref100/uniref100.fasta',
+		'db/uniref/uniref90/uniref90.fasta',
+		'db/uniref/uniref50/uniref50.fasta'
+	],
+        'dirs': ['db','blastdb_v4','blastdb_v5','diamond']
+}
+
+modextrapaths = {'BIODB':'db',
+	'BLASTDB': ['blastdb_v4','blastdb_v5'],
+	'DIAMONDDB': 'diamond',
+}
+
+keeppreviousinstall = True
+
+moduleclass = 'data'
+
+modloadmsg = "%(name)s/%(version)s database and indexes are located at %(installdir)s/\n"
+
+##Helpful documentation
+#https://embl.service-now.com/kb?id=kb_article_view&sysparm_article=KB0011060
diff --git a/easyconfigs/u/uniprot/uniprot_download.sh b/easyconfigs/u/uniprot/uniprot_download.sh
@@ -1,5 +1,5 @@
 # ----------------Load Modules--------------------
-module load globus-cli/3.18.0-IGB-gcc-8.2.0-Python-3.7.2
+module load globus-cli/3.30.1-IGB-gcc-8.2.0-Python-3.10.1
 
 # ----------------Commands------------------------
 
@@ -15,7 +15,7 @@ FASTA_DIR=${MIRROR_DIR}/db
 
 echo "Downloading Files: `date "+%Y-%m-%d %k:%M:%S"`"
 mkdir -p ${FASTA_DIR}
-globus transfer -r --exclude "reference_proteomes" --exclude "taxonomic_divisions" "47772002-3e5b-4fd3-b97c-18cee38d6df2:/pub/databases/uniprot/current_release/" "1ccc563b-0542-44e5-a13c-fc4b00281b72:${FASTA_DIR}/"
+globus transfer -r --exclude "reference_proteomes" --exclude "taxonomic_divisions" "47772002-3e5b-4fd3-b97c-18cee38d6df2:/pub/databases/uniprot/current_release/" "4a467fda-f559-4fc3-b54a-e2842f439e06:${FASTA_DIR}/"