Repository 'data_manager_build_kraken2_database'
hg clone https://testtoolshed.g2.bx.psu.edu/repos/dfornika/data_manager_build_kraken2_database

Changeset 34:8d8ebf8367c8 (2019-05-01)
Previous changeset 33:eb123ede1812 (2019-04-30) Next changeset 35:40b2c0a4abd5 (2019-05-01)
Commit message:
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_kraken2_database/ commit 5629d9d7d9836d32b1e501277e213ef213de3e76-dirty
modified:
data_manager/kraken2_build_database.py
b
diff -r eb123ede1812 -r 8d8ebf8367c8 data_manager/kraken2_build_database.py
--- a/data_manager/kraken2_build_database.py Tue Apr 30 15:04:10 2019 -0400
+++ b/data_manager/kraken2_build_database.py Wed May 01 16:46:36 2019 -0400
[
@@ -9,9 +9,25 @@
 import os
 import subprocess
 
+from enum import Enum
 
 DATA_TABLE_NAME = "kraken2_databases"
 
+class KrakenDatabaseTypes(Enum):
+    standard = 'standard'
+    minikraken = 'minikraken'
+    special = 'special'
+    custom = 'custom'
+
+    def __str__(self):
+        return self.value
+
+class Minikraken2Versions(Enum):
+    v1 = 'v1'
+    v2 = 'v2'
+
+    def __str__(self):
+        return self.value
 
 def kraken2_build_standard(data_manager_dict, kraken2_args, target_directory, data_table_name=DATA_TABLE_NAME):
     now = datetime.datetime.utcnow().strftime("%Y-%m-%dT%H%M%SZ")
@@ -63,6 +79,43 @@
     _add_data_table_entry(data_manager_dict, data_table_entry)
 
 
+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")
+
+    database_value = "_".join([
+        now,
+        "minikraken2",
+        minikraken2_version,
+        "8GB",
+    ])
+
+    database_name = " ".join([
+        "Minikraken2",
+        minikraken2_version,
+        "(Created:",
+        now + ")"
+    ])
+
+    # download the minikraken2 data
+    src = urlopen(
+        'ftp://ftp.ccb.jhu.edu/pub/data/kraken2_dbs/minikraken2_%s_8GB_201904_UPDATE.tgz'
+        % minikraken2_version
+    )
+    with open('tmp_data.tar.gz', 'wb') as dst:
+        shutil.copyfileobj(src, dst)
+    # unpack the downloaded archive to the target directory
+    with tarfile.open('tmp_data.tar.gz', 'r:gz') as fh:
+        fh.extractall(target_directory)
+
+    data_table_entry = {
+        "value": database_value,
+        "name": database_name,
+        "path": database_value,
+    }
+
+    _add_data_table_entry(data_manager_dict, data_table_entry)
+
 def _add_data_table_entry(data_manager_dict, data_table_entry, data_table_name=DATA_TABLE_NAME):
     data_manager_dict['data_tables'] = data_manager_dict.get( 'data_tables', {} )
     data_manager_dict['data_tables'][data_table_name] = data_manager_dict['data_tables'].get( data_table_name, [] )
@@ -73,18 +126,14 @@
 def main():
     parser = argparse.ArgumentParser()
     parser.add_argument('data_manager_json')
-    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( '-t', '--threads', dest='threads', default=1, help='threads' )
+    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('--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' )
     args = parser.parse_args()
 
-    kraken2_args = {
-        "kmer_len": args.kmer_len,
-        "minimizer_len": args.minimizer_len,
-        "minimizer_spaces": args.minimizer_spaces,
-        "threads": args.threads,
-    }
 
     data_manager_input = json.loads(open(args.data_manager_json).read())
 
@@ -100,11 +149,26 @@
 
     data_manager_output = {}
 
-    kraken2_build_standard(
-        data_manager_output,
-        kraken2_args,
-        target_directory,
-    )
+    if args.database_type == 'standard':
+        kraken2_args = {
+            "kmer_len": args.kmer_len,
+            "minimizer_len": args.minimizer_len,
+            "minimizer_spaces": args.minimizer_spaces,
+            "threads": args.threads,
+        }
+        kraken2_build_standard(
+            data_manager_output,
+            kraken2_args,
+            target_directory,
+        )
+    elif args.database_type == 'minikraken':
+        kraken2_build_minikraken(
+            data_manager_output,
+            args.minikraken2_version,
+            target_directory
+        )
+    else:
+        sys.exit("Invalid database type")
 
     open(args.data_manager_json, 'w').write(json.dumps(data_manager_output))