diff data_manager/kraken2_build_database.py @ 0:bd47b9f87d67 draft default tip

"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_kraken2_database/ commit 68cd9a8ae50c5dfe6b667062a5172010511bcaff-dirty"

--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/data_manager/kraken2_build_database.py	Tue Dec 01 16:07:40 2020 +0000
@@ -0,0 +1,382 @@
+#!/usr/bin/env python
+
+from __future__ import print_function
+
+import argparse
+import datetime
+import errno
+import json
+import os
+import shutil
+import subprocess
+import sys
+import tarfile
+from enum import Enum
+
+try:
+    # Python3
+    from urllib.request import urlopen
+except ImportError:
+    from urllib2 import urlopen
+
+
+DATA_TABLE_NAME = "kraken2_databases"
+
+
+class KrakenDatabaseTypes(Enum):
+    standard = 'standard'
+    minikraken = 'minikraken'
+    special = 'special'
+    custom = 'custom'
+
+    def __str__(self):
+        return self.value
+
+
+class SpecialDatabaseTypes(Enum):
+    rdp = 'rdp'
+    greengenes = 'greengenes'
+    silva = 'silva'
+
+    def __str__(self):
+        return self.value
+
+
+class Minikraken2Versions(Enum):
+    v1 = 'v1'
+    v2 = 'v2'
+
+    def __str__(self):
+        return self.value
+
+class Minikraken2Releases(Enum):
+    March_2020 = 'March_2020'
+    April_2019 = 'April_2019'
+
+    def __str__(self):
+        return self.value
+
+
+def kraken2_build_standard(kraken2_args, target_directory, data_table_name=DATA_TABLE_NAME):
+    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"]),
+        "load-factor", str(kraken2_args["load_factor"]),
+    ])
+
+    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"]) + ")",
+        "load-factor", str(kraken2_args["load_factor"]),
+    ])
+
+    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"]),
+        '--load-factor', str(kraken2_args["load_factor"]),
+        '--db', database_path
+    ]
+
+    subprocess.check_call(['kraken2-build'] + args, cwd=target_directory)
+
+    if kraken2_args["clean"]:
+        args = [
+            '--threads', str(kraken2_args["threads"]),
+            '--clean',
+            '--db', database_path
+        ]
+
+        subprocess.check_call(['kraken2-build'] + args, cwd=target_directory)
+
+    data_table_entry = {
+        'data_tables': {
+            data_table_name: [
+                {
+                    "value": database_value,
+                    "name": database_name,
+                    "path": database_path,
+                }
+            ]
+        }
+    }
+
+    return data_table_entry
+
+
+def kraken2_build_minikraken(minikraken2_version, minikraken2_release, target_directory, data_table_name=DATA_TABLE_NAME):
+
+    now = datetime.datetime.utcnow().strftime("%Y-%m-%dT%H%M%SZ")
+
+    value_parts = [now, "minikraken2", minikraken2_release, "8GB"]
+    name_parts = ["Minikraken2", minikraken2_release, "8GB", "(Created: %s)" % now]
+
+    if minikraken2_release == 'April_2019':
+        value_parts.insert(3, minikraken2_version)
+        name_parts.insert(2, minikraken2_version)
+        src = urlopen(
+            'ftp://ftp.ccb.jhu.edu/pub/data/kraken2_dbs/old/minikraken2_%s_8GB_201904.tgz'
+            % minikraken2_version
+        )
+    else:
+        src = urlopen('ftp://ftp.ccb.jhu.edu/pub/data/kraken2_dbs/minikraken_8GB_202003.tgz')
+
+    database_value = "_".join(value_parts)
+
+    database_name = " ".join(name_parts)
+
+    database_path = database_value
+
+    # download the minikraken2 data
+    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:
+        for member in fh.getmembers():
+            if member.isreg():
+                member.name = os.path.basename(member.name)
+                fh.extract(member, os.path.join(target_directory, database_path))
+
+    data_table_entry = {
+        'data_tables': {
+            data_table_name: [
+                {
+                    "value": database_value,
+                    "name": database_name,
+                    "path": database_path,
+                }
+            ]
+        }
+    }
+
+    return data_table_entry
+
+
+def kraken2_build_special(kraken2_args, 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_database_type"],
+        "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([
+        special_database_names[kraken2_args["special_database_type"]],
+        "(Created:",
+        now + ",",
+        "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
+
+    args = [
+        '--threads', str(kraken2_args["threads"]),
+        '--special', kraken2_args["special_database_type"],
+        '--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
+    ]
+
+    subprocess.check_call(['kraken2-build'] + args, cwd=target_directory)
+
+    if kraken2_args["clean"]:
+        args = [
+            '--threads', str(kraken2_args["threads"]),
+            '--clean',
+            '--db', database_path
+        ]
+
+        subprocess.check_call(['kraken2-build'] + args, cwd=target_directory)
+
+    data_table_entry = {
+        'data_tables': {
+            data_table_name: [
+                {
+                    "value": database_value,
+                    "name": database_name,
+                    "path": database_path,
+                }
+            ]
+        }
+    }
+
+    return data_table_entry
+
+
+def kraken2_build_custom(kraken2_args, custom_database_name, target_directory, data_table_name=DATA_TABLE_NAME):
+
+    args = [
+        '--threads', str(kraken2_args["threads"]),
+        '--download-taxonomy',
+        '--db', custom_database_name,
+    ]
+
+    if kraken2_args['skip_maps']:
+        args.append('--skip-maps')
+
+    subprocess.check_call(['kraken2-build'] + args, cwd=target_directory)
+
+    args = [
+        '--threads', str(kraken2_args["threads"]),
+        '--add-to-library', kraken2_args["custom_fasta"],
+        '--db', custom_database_name
+    ]
+
+    subprocess.check_call(['kraken2-build'] + args, cwd=target_directory)
+
+    args = [
+        '--threads', str(kraken2_args["threads"]),
+        '--build',
+        '--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
+    ]
+
+    subprocess.check_call(['kraken2-build'] + args, cwd=target_directory)
+
+    if kraken2_args["clean"]:
+        args = [
+            '--threads', str(kraken2_args["threads"]),
+            '--clean',
+            '--db', custom_database_name
+        ]
+
+        subprocess.check_call(['kraken2-build'] + args, cwd=target_directory)
+
+    data_table_entry = {
+        'data_tables': {
+            data_table_name: [
+                {
+                    "value": custom_database_name,
+                    "name": custom_database_name,
+                    "path": custom_database_name
+                }
+            ]
+        }
+    }
+
+    return data_table_entry
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument('data_manager_json')
+    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) and the Mar2019 release')
+    parser.add_argument('--minikraken2-release', dest='minikraken2_release', type=Minikraken2Releases, choices=list(Minikraken2Releases), help='MiniKraken2 release (only applies to --database-type minikraken)')
+    parser.add_argument('--special-database-type', dest='special_database_type', type=SpecialDatabaseTypes, choices=list(SpecialDatabaseTypes), help='type of special database to build (only applies to --database-type special)')
+    parser.add_argument('--custom-fasta', dest='custom_fasta', help='fasta file for custom database (only applies to --database-type custom)')
+    parser.add_argument('--custom-database-name', dest='custom_database_name', help='Name for custom database (only applies to --database-type custom)')
+    parser.add_argument('--skip-maps', dest='skip_maps', action='store_true', help='')
+    parser.add_argument('--clean', dest='clean', action='store_true', help='Clean up extra files')
+    args = parser.parse_args()
+
+    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)
+    except OSError as exc:
+        if exc.errno == errno.EEXIST and os.path.isdir(target_directory):
+            pass
+        else:
+            raise
+
+    data_manager_output = {}
+
+    if str(args.database_type) == 'standard':
+        kraken2_args = {
+            "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,
+        }
+        data_manager_output = kraken2_build_standard(
+            kraken2_args,
+            target_directory,
+        )
+    elif str(args.database_type) == 'minikraken':
+        data_manager_output = kraken2_build_minikraken(
+            str(args.minikraken2_version),
+            str(args.minikraken2_release),
+            target_directory
+        )
+    elif str(args.database_type) == 'special':
+        kraken2_args = {
+            "special_database_type": str(args.special_database_type),
+            "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,
+        }
+        data_manager_output = kraken2_build_special(
+            kraken2_args,
+            target_directory,
+        )
+    elif str(args.database_type) == 'custom':
+        kraken2_args = {
+            "custom_fasta": args.custom_fasta,
+            "skip_maps": args.skip_maps,
+            "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,
+        }
+        data_manager_output = kraken2_build_custom(
+            kraken2_args,
+            args.custom_database_name,
+            target_directory,
+        )
+    else:
+        sys.exit("Invalid database type")
+
+    with open(args.data_manager_json, 'w') as fh:
+        json.dump(data_manager_output, fh, sort_keys=True)
+
+
+if __name__ == "__main__":
+    main()