# HG changeset patch # User dfornika # Date 1635978868 0 # Node ID 80e778af6c6c7dc3186b21384a84152ccd0111bb # Parent 38c99b2942645ad6951b38f2463a747c05e6fd81 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_kraken2_database/ commit 57b36cfbdd1663aef43d03b76e37364cc5bdeef3-dirty" diff -r 38c99b294264 -r 80e778af6c6c data_manager/kraken2_build_database.py --- a/data_manager/kraken2_build_database.py Wed Oct 23 14:18:06 2019 -0400 +++ b/data_manager/kraken2_build_database.py Wed Nov 03 22:34:28 2021 +0000 @@ -59,6 +59,7 @@ "kmer-len", str(kraken2_args["kmer_len"]), "minimizer-len", str(kraken2_args["minimizer_len"]), "minimizer-spaces", str(kraken2_args["minimizer_spaces"]), + "load-factor", str(kraken2_args["load_factor"]), ]) database_name = " ".join([ @@ -68,6 +69,7 @@ "kmer-len=" + str(kraken2_args["kmer_len"]) + ",", "minimizer-len=" + str(kraken2_args["minimizer_len"]) + ",", "minimizer-spaces=" + str(kraken2_args["minimizer_spaces"]) + ")", + "load-factor", str(kraken2_args["load_factor"]), ]) database_path = database_value @@ -78,6 +80,7 @@ '--kmer-len', str(kraken2_args["kmer_len"]), '--minimizer-len', str(kraken2_args["minimizer_len"]), '--minimizer-spaces', str(kraken2_args["minimizer_spaces"]), + '--load-factor', str(kraken2_args["load_factor"]), '--db', database_path ] @@ -129,7 +132,7 @@ # download the minikraken2 data src = urlopen( - 'ftp://ftp.ccb.jhu.edu/pub/data/kraken2_dbs/minikraken2_%s_8GB_201904_UPDATE.tgz' + 'https://genome-idx.s3.amazonaws.com/kraken/minikraken2_%s_8GB_201904.tgz' % minikraken2_version ) with open('tmp_data.tar.gz', 'wb') as dst: @@ -172,6 +175,7 @@ "kmer-len", str(kraken2_args["kmer_len"]), "minimizer-len", str(kraken2_args["minimizer_len"]), "minimizer-spaces", str(kraken2_args["minimizer_spaces"]), + "load-factor", str(kraken2_args["load_factor"]), ]) database_name = " ".join([ @@ -181,6 +185,7 @@ "kmer-len=" + str(kraken2_args["kmer_len"]) + ",", "minimizer-len=" + str(kraken2_args["minimizer_len"]) + ",", "minimizer-spaces=" + str(kraken2_args["minimizer_spaces"]) + ")", + "load-factor=" + str(kraken2_args["load_factor"]) + ")", ]) database_path = database_value @@ -191,6 +196,7 @@ '--kmer-len', str(kraken2_args["kmer_len"]), '--minimizer-len', str(kraken2_args["minimizer_len"]), '--minimizer-spaces', str(kraken2_args["minimizer_spaces"]), + '--load-factor', str(kraken2_args["load_factor"]), '--db', database_path ] @@ -247,6 +253,7 @@ '--kmer-len', str(kraken2_args["kmer_len"]), '--minimizer-len', str(kraken2_args["minimizer_len"]), '--minimizer-spaces', str(kraken2_args["minimizer_spaces"]), + '--load-factor', str(kraken2_args["load_factor"]), '--db', custom_database_name ] @@ -282,6 +289,7 @@ parser.add_argument('--kmer-len', dest='kmer_len', type=int, default=35, help='kmer length') parser.add_argument('--minimizer-len', dest='minimizer_len', type=int, default=31, help='minimizer length') parser.add_argument('--minimizer-spaces', dest='minimizer_spaces', default=6, help='minimizer spaces') + parser.add_argument('--load-factor', dest='load_factor', type=float, default=0.7, help='load factor') parser.add_argument('--threads', dest='threads', default=1, help='threads') parser.add_argument('--database-type', dest='database_type', type=KrakenDatabaseTypes, choices=list(KrakenDatabaseTypes), required=True, help='type of kraken database to build') parser.add_argument('--minikraken2-version', dest='minikraken2_version', type=Minikraken2Versions, choices=list(Minikraken2Versions), help='MiniKraken2 version (only applies to --database-type minikraken)') @@ -292,14 +300,15 @@ parser.add_argument('--clean', dest='clean', action='store_true', help='Clean up extra files') args = parser.parse_args() - data_manager_input = json.loads(open(args.data_manager_json).read()) + with open(args.data_manager_json) as fh: + data_manager_input = json.load(fh) target_directory = data_manager_input['output_data'][0]['extra_files_path'] try: - os.mkdir( target_directory ) + os.mkdir(target_directory) except OSError as exc: - if exc.errno == errno.EEXIST and os.path.isdir( target_directory ): + if exc.errno == errno.EEXIST and os.path.isdir(target_directory): pass else: raise @@ -311,6 +320,7 @@ "kmer_len": args.kmer_len, "minimizer_len": args.minimizer_len, "minimizer_spaces": args.minimizer_spaces, + "load_factor": args.load_factor, "threads": args.threads, "clean": args.clean, } @@ -329,6 +339,7 @@ "kmer_len": args.kmer_len, "minimizer_len": args.minimizer_len, "minimizer_spaces": args.minimizer_spaces, + "load_factor": args.load_factor, "threads": args.threads, "clean": args.clean, } @@ -343,6 +354,7 @@ "kmer_len": args.kmer_len, "minimizer_len": args.minimizer_len, "minimizer_spaces": args.minimizer_spaces, + "load_factor": args.load_factor, "threads": args.threads, "clean": args.clean, } @@ -354,7 +366,8 @@ else: sys.exit("Invalid database type") - open(args.data_manager_json, 'w').write(json.dumps(data_manager_output)) + with open(args.data_manager_json, 'w') as fh: + json.dump(data_manager_output, fh, sort_keys=True) if __name__ == "__main__": diff -r 38c99b294264 -r 80e778af6c6c data_manager/kraken2_build_database.xml --- a/data_manager/kraken2_build_database.xml Wed Oct 23 14:18:06 2019 -0400 +++ b/data_manager/kraken2_build_database.xml Wed Nov 03 22:34:28 2021 +0000 @@ -1,16 +1,19 @@ - + + 2.1.1 + database builder - kraken2 + kraken2 + python kraken2 -version | head -n 1 | awk '{print $NF}' @@ -25,6 +28,7 @@ --kmer-len ${database_type.kmer_len} --minimizer-len ${database_type.minimizer_len} --minimizer-spaces ${database_type.minimizer_spaces} + --load-factor ${database_type.load_factor} ${database_type.clean} #else if $database_type.database_type == "special" --threads \${GALAXY_SLOTS:-1} @@ -32,6 +36,7 @@ --kmer-len ${database_type.kmer_len} --minimizer-len ${database_type.minimizer_len} --minimizer-spaces ${database_type.minimizer_spaces} + --load-factor ${database_type.load_factor} ${database_type.clean} #else if $database_type.database_type == "custom" --threads \${GALAXY_SLOTS:-1} @@ -41,6 +46,7 @@ --kmer-len ${database_type.kmer_len} --minimizer-len ${database_type.minimizer_len} --minimizer-spaces ${database_type.minimizer_spaces} + --load-factor ${database_type.load_factor} ${database_type.clean} #end if ]]> diff -r 38c99b294264 -r 80e778af6c6c data_manager_conf.xml --- a/data_manager_conf.xml Wed Oct 23 14:18:06 2019 -0400 +++ b/data_manager_conf.xml Wed Nov 03 22:34:28 2021 +0000 @@ -1,5 +1,5 @@ - + diff -r 38c99b294264 -r 80e778af6c6c test-data/kraken2_custom_data_manager.json --- a/test-data/kraken2_custom_data_manager.json Wed Oct 23 14:18:06 2019 -0400 +++ b/test-data/kraken2_custom_data_manager.json Wed Nov 03 22:34:28 2021 +0000 @@ -1,1 +1,1 @@ -{"data_tables": {"kraken2_databases": [{"path": "database", "name": "database", "value": "database"}]}} \ No newline at end of file +{"data_tables": {"kraken2_databases": [{"name": "database", "path": "database", "value": "database"}]}} \ No newline at end of file