comparison data_manager/kraken2_build_database.py @ 12:90b4d4f0a3a4 draft

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/data_managers/data_manager_build_kraken2_database/ commit 9835da32741d05d129a1a44835f66e32713770ad

11:1e34d2e3d285

#!/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

    }

    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 = [

        '--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
                }
            ]
        }
    }

    parser.add_argument('--prebuilt-db', dest='prebuilt_db', type=StandardPrebuiltSizes, choices=list(StandardPrebuiltSizes), help='Prebuilt database to download. Only applies to --database-type standard_prebuilt.')
    parser.add_argument('--prebuilt-date', dest='prebuilt_date', help='Database build date (YYYY-MM-DD). Only applies to --database-type standard_prebuilt.')
    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:

            "clean": args.clean,
        }
        data_manager_output = kraken2_build_custom(
            kraken2_args,
            args.custom_database_name,
            target_directory,
        )
    else:
        sys.exit("Invalid database type")

12:90b4d4f0a3a4

#!/usr/bin/env python

import argparse
import datetime
import errno
import json
import os
import re
import shutil
import subprocess
import sys
import tarfile
from enum import Enum

    }

    return data_table_entry


def kraken2_build_custom(kraken2_args, custom_database_name, custom_source_info, target_directory, data_table_name=DATA_TABLE_NAME):
    now = datetime.datetime.utcnow().strftime("%Y-%m-%dT%H%M%SZ")

    database_value = "_".join([
        now,
        re.sub(r'[^\w_.-]+', '_', custom_database_name).strip('_'),
        "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([
        custom_database_name,
        "(" + custom_source_info + ",",
        "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"]),
        '--download-taxonomy',
        '--db', database_path,
    ]

    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', database_path,
    ]

    subprocess.check_call(['kraken2-build'] + args, cwd=target_directory)

    args = [

        '--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', 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,
                }
            ]
        }
    }

    parser.add_argument('--prebuilt-db', dest='prebuilt_db', type=StandardPrebuiltSizes, choices=list(StandardPrebuiltSizes), help='Prebuilt database to download. Only applies to --database-type standard_prebuilt.')
    parser.add_argument('--prebuilt-date', dest='prebuilt_date', help='Database build date (YYYY-MM-DD). Only applies to --database-type standard_prebuilt.')
    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('--custom-source-info', dest='custom_source_info', help='Description of how this build has been sourced (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:

            "clean": args.clean,
        }
        data_manager_output = kraken2_build_custom(
            kraken2_args,
            args.custom_database_name,
            args.custom_source_info,
            target_directory,
        )
    else:
        sys.exit("Invalid database type")