# HG changeset patch # User dfornika # Date 1551816665 18000 # Node ID 039a65ff445d7a751a56c1b533bd72e01f5118dc # Parent 1dc93ae264e699f7b5c857edb24d13b22a64de25 planemo upload for repository https://github.com/dfornika/galaxytools/tree/master/data_managers/data_manager_build_kraken2_database/ commit 70494d589aa1990618ebc7a895e91477d41c3203-dirty diff -r 1dc93ae264e6 -r 039a65ff445d data_manager/kraken2_build_minikraken.py --- a/data_manager/kraken2_build_minikraken.py Tue Mar 05 14:33:23 2019 -0500 +++ b/data_manager/kraken2_build_minikraken.py Tue Mar 05 15:11:05 2019 -0500 @@ -21,53 +21,37 @@ print("Error building database.", file=sys.stderr) sys.exit( return_code ) -def kraken2_build_standard(data_manager_dict, kraken2_args, target_directory, data_table_name=DATA_TABLE_NAME): +def kraken2_build_minikraken(data_manager_dict, minikraken2_version, target_directory, data_table_name=DATA_TABLE_NAME): now = datetime.datetime.utcnow().strftime("%Y-%m-%dT%H%M%SZ") - - special_database_names = { - "rdp": "RDP", - "greengenes": "Greengenes", - "silva": "Silva", - } database_value = "_".join([ now, - kraken2_args["special"], - "kmer-len", str(kraken2_args["kmer_len"]), - "minimizer-len", str(kraken2_args["minimizer_len"]), - "minimizer-spaces", str(kraken2_args["minimizer_spaces"]), + "minikraken2", + minikraken2_version, + "8GB", ]) database_name = " ".join([ - special_database_names[kraken2_args["special"]], + "Minikraken2", + minikraken2_version, "(Created:", - now + "," - "kmer-len=" + str(kraken2_args["kmer_len"]) + ",", - "minimizer-len=" + str(kraken2_args["minimizer_len"]) + ",", - "minimizer-spaces=" + str(kraken2_args["minimizer_spaces"]) + ")", + now + ")" ]) database_path = database_value args = [ - '--threads', str(kraken2_args["threads"]), - '--special', kraken2_args["special"], - '--kmer-len', str(kraken2_args["kmer_len"]), - '--minimizer-len', str(kraken2_args["minimizer_len"]), - '--minimizer-spaces', str(kraken2_args["minimizer_spaces"]), - '--db', database_path + 'https://ccb.jhu.edu/software/kraken2/dl/minikraken2_' + minikraken2_version + '_8GB.tgz' ] - run(['kraken2-build'] + args, target_directory) + run(['wget'] + args, target_directory) args = [ - '--threads', str(kraken2_args["threads"]), - '--clean', - '--db', database_path + '-xvzf', 'minikraken2_' + minikraken2_version + '_8GB.tgz' ] - run(['kraken2-build'] + args, target_directory) + run(['tar'] + args, target_directory) data_table_entry = { "value": database_value, @@ -88,20 +72,10 @@ def main(): parser = argparse.ArgumentParser() parser.add_argument('data_manager_json') - parser.add_argument( '-b', '--db-type', dest='db_type', help='database type (one of: silva, rdp, greengenes)' ) - parser.add_argument( '-k', '--kmer-len', dest='kmer_len', type=int, default=35, help='kmer length' ) - parser.add_argument( '-m', '--minimizer-len', dest='minimizer_len', type=int, default=31, help='minimizer length' ) - parser.add_argument( '-s', '--minimizer-spaces', dest='minimizer_spaces', default=6, help='minimizer spaces' ) + parser.add_argument( '-v', '--minikraken2-version', dest='minikraken2_version', default="v2", help='MiniKraken2 version (v1 or v2)' ) parser.add_argument( '-t', '--threads', dest='threads', default=1, help='threads' ) - args = parser.parse_args() - kraken2_args = { - "special": args.db_type, - "kmer_len": args.kmer_len, - "minimizer_len": args.minimizer_len, - "minimizer_spaces": args.minimizer_spaces, - "threads": args.threads, - } + args = parser.parse_args() data_manager_input = json.loads(open(args.data_manager_json).read()) @@ -117,9 +91,9 @@ data_manager_output = {} - kraken2_build_standard( + kraken2_build_minikraken( data_manager_output, - kraken2_args, + args.minikraken2_version, target_directory, ) diff -r 1dc93ae264e6 -r 039a65ff445d data_manager/kraken2_build_minikraken.xml --- a/data_manager/kraken2_build_minikraken.xml Tue Mar 05 14:33:23 2019 -0500 +++ b/data_manager/kraken2_build_minikraken.xml Tue Mar 05 15:11:05 2019 -0500 @@ -1,5 +1,5 @@ - + database builder kraken2 @@ -7,24 +7,17 @@ kraken2 -version | head -n 1 | awk '{print $NF}' - - - - + + + - - - diff -r 1dc93ae264e6 -r 039a65ff445d data_manager/kraken2_build_special.py --- a/data_manager/kraken2_build_special.py Tue Mar 05 14:33:23 2019 -0500 +++ b/data_manager/kraken2_build_special.py Tue Mar 05 15:11:05 2019 -0500 @@ -42,7 +42,7 @@ database_name = " ".join([ special_database_names[kraken2_args["special"]], "(Created:", - now + "," + now + ",", "kmer-len=" + str(kraken2_args["kmer_len"]) + ",", "minimizer-len=" + str(kraken2_args["minimizer_len"]) + ",", "minimizer-spaces=" + str(kraken2_args["minimizer_spaces"]) + ")", diff -r 1dc93ae264e6 -r 039a65ff445d data_manager/kraken2_build_standard.py --- a/data_manager/kraken2_build_standard.py Tue Mar 05 14:33:23 2019 -0500 +++ b/data_manager/kraken2_build_standard.py Tue Mar 05 15:11:05 2019 -0500 @@ -22,21 +22,34 @@ sys.exit( return_code ) def kraken2_build_standard(data_manager_dict, kraken2_args, target_directory, data_table_name=DATA_TABLE_NAME): - today = datetime.date.today().isoformat() - database_name = "_".join([ - today, + now = datetime.datetime.utcnow().strftime("%Y-%m-%dT%H%M%SZ") + + database_value = "_".join([ + now, "standard", - "kmer-len=" + str(kraken2_args["kmer_len"]), - "minimizer-len=" + str(kraken2_args["minimizer_len"]), - "minimizer-spaces=" + str(kraken2_args["minimizer_spaces"]), + "kmer-len" + str(kraken2_args["kmer_len"]), + "minimizer-len" + str(kraken2_args["minimizer_len"]), + "minimizer-spaces" + str(kraken2_args["minimizer_spaces"]), ]) + + database_name = " ".join([ + "Standard", + "(Created:", + now + ",", + "kmer-len=" + str(kraken2_args["kmer_len"]) + ",", + "minimizer-len=" + str(kraken2_args["minimizer_len"]) + ",", + "minimizer-spaces=" + str(kraken2_args["minimizer_spaces"]) + ")", + ]) + + database_path = database_value + args = [ '--threads', str(kraken2_args["threads"]), '--standard', '--kmer-len', str(kraken2_args["kmer_len"]), '--minimizer-len', str(kraken2_args["minimizer_len"]), '--minimizer-spaces', str(kraken2_args["minimizer_spaces"]), - '--db', database_name + '--db', database_path ] run(['kraken2-build'] + args, target_directory) @@ -44,15 +57,15 @@ args = [ '--threads', str(kraken2_args["threads"]), '--clean', - '--db', database_name + '--db', database_path ] run(['kraken2-build'] + args, target_directory) data_table_entry = { - "value": database_name, + "value": database_value, "name": database_name, - "path": database_name + "path": database_path, } _add_data_table_entry(data_manager_dict, data_table_entry) diff -r 1dc93ae264e6 -r 039a65ff445d data_manager_conf.xml --- a/data_manager_conf.xml Tue Mar 05 14:33:23 2019 -0500 +++ b/data_manager_conf.xml Tue Mar 05 15:11:05 2019 -0500 @@ -47,4 +47,20 @@ + + + + + + + + ${path} + kraken2_databases/${path} + + ${GALAXY_DATA_MANAGER_DATA_PATH}/kraken2_databases/${path} + abspath + + + +