def setUp(self):
blast_databases_repositories = AMRDatabasesManager.create_default_manager(
).get_database_repos()
self.resfinder_dir = blast_databases_repositories.get_repo_dir(
'resfinder')
self.pointfinder_dir = blast_databases_repositories.get_repo_dir(
'pointfinder')
self.plasmidfinder_dir = blast_databases_repositories.get_repo_dir(
'plasmidfinder')
self.resfinder_database = ResfinderBlastDatabase(self.resfinder_dir)
self.resfinder_drug_table = ARGDrugTableResfinder()
self.pointfinder_drug_table = ARGDrugTablePointfinder()
self.plasmidfinder_database = PlasmidfinderBlastDatabase(
self.plasmidfinder_dir)
self.pointfinder_database = None
self.blast_out = tempfile.TemporaryDirectory()
self.blast_handler = JobHandler(
{
'resfinder': self.resfinder_database,
'pointfinder': self.pointfinder_database,
'plasmidfinder': self.plasmidfinder_database
}, 2, self.blast_out.name)
self.outdir = tempfile.TemporaryDirectory()
self.amr_detection = AMRDetectionResistance(
self.resfinder_database,
self.resfinder_drug_table,
self.blast_handler,
self.pointfinder_drug_table,
self.pointfinder_database,
output_dir=self.outdir.name)
self.test_data_dir = path.join(path.dirname(__file__), '..', 'data')
def run(self, args):
super(Build, self).run(args)
if path.exists(args.destination):
if args.destination == self._default_dir:
raise CommandParseException("Error, default destination [" +
args.destination +
"] already exists",
self._root_arg_parser,
print_help=True)
else:
raise CommandParseException(
"Error, destination [" + args.destination +
"] already exists", self._root_arg_parser)
else:
mkdir(args.destination)
if args.destination == AMRDatabasesManager.get_default_database_directory(
):
database_repos = AMRDatabasesManager.create_default_manager(
).get_database_repos()
else:
database_repos = AMRDatabasesManager(
args.destination).get_database_repos()
database_repos.build({
'resfinder': args.resfinder_commit,
'pointfinder': args.pointfinder_commit
})
if not AMRDatabasesManager.is_database_repos_default_commits(
database_repos):
logger.warning(
"Built non-default ResFinder/PointFinder database version. This may lead to "
+
"differences in the detected AMR genes depending on how the database files are structured."
)
def run(self, args):
super(Update, self).run(args)
if len(args.directories) == 0:
if not args.update_default:
raise CommandParseException("Must pass at least one directory to update, or use '--update-default'",
self._root_arg_parser,
print_help=True)
else:
try:
database_repos = AMRDatabasesManager.create_default_manager().get_database_repos(
force_use_git=True)
database_repos.update(
{'resfinder': args.resfinder_commit, 'pointfinder': args.pointfinder_commit,
'plasmidfinder': args.plasmidfinder_commit})
if not AMRDatabasesManager.is_database_repos_default_commits(database_repos):
logger.warning(
"Updated to non-default ResFinder/PointFinder/PlasmidFinder database version. This may lead to " +
"differences in the detected AMR genes depending on how the database files are structured.")
except DatabaseErrorException as e:
logger.error(
"Could not update default database. Please try restoring with 'staramr db restore-default'")
raise e
else:
for directory in args.directories:
database_repos = AMRDatabasesManager(directory).get_database_repos()
database_repos.update({'resfinder': args.resfinder_commit, 'pointfinder': args.pointfinder_commit,
'plasmidfinder': args.plasmidfinder_commit})
if not AMRDatabasesManager.is_database_repos_default_commits(database_repos):
logger.warning(
"Updated to non-default ResFinder/PointFinder/PlasmidFinder database version [%s]. This may lead to " +
"differences in the detected AMR genes depending on how the database files are structured.",
directory)
def testIsHandlerDefaultCommitsTrue(self):
# Setup default database
self.databases_manager.setup_default()
database_handler = self.databases_manager.get_database_handler()
self.assertTrue(
AMRDatabasesManager.is_handler_default_commits(database_handler),
"Database is not default")
def testIsHandlerDefaultCommitsTrue(self):
# Setup default database
self.databases_manager.setup_default()
blast_database_repos = self.databases_manager.get_database_repos()
self.assertTrue(
AMRDatabasesManager.is_database_repos_default_commits(
blast_database_repos), "Database is not default")
def testIsHandlerDefaultCommitsFalse(self):
# Setup database
blast_database_repos = self.databases_manager.get_database_repos(
force_use_git=True)
blast_database_repos.update(
{'resfinder': 'dc33e2f9ec2c420f99f77c5c33ae3faa79c999f2'})
self.assertFalse(
AMRDatabasesManager.is_database_repos_default_commits(
blast_database_repos), "Database is default")
def testIsHandlerDefaultCommitsFalse(self):
# Setup database
database_handler = self.databases_manager.get_database_handler(
force_use_git=True)
database_handler.update(
resfinder_commit='dc33e2f9ec2c420f99f77c5c33ae3faa79c999f2')
self.assertFalse(
AMRDatabasesManager.is_handler_default_commits(database_handler),
"Database is default")
def run(self, args):
super(RestoreDefault, self).run(args)
database_manager = AMRDatabasesManager.create_default_manager()
if not args.force:
response = self._confirm_restore()
else:
response = True
if response:
database_manager.restore_default()
def _setup_args(self, arg_parser):
name = self._script_name
default_dir = AMRDatabasesManager.get_default_database_directory()
epilog = ("Example:\n"
"\t" + name + " info\n"
"\t\tPrints information about the default database in " + default_dir + "\n\n" +
"\t" + name + " info databases\n" +
"\t\tPrints information on the database stored in databases/")
arg_parser = self._subparser.add_parser('info',
epilog=epilog,
formatter_class=argparse.RawTextHelpFormatter,
help='Prints information on databases in the given directories.')
arg_parser.add_argument('directories', nargs='*')
return arg_parser
def _setup_args(self, arg_parser):
name = self._script_name
self._default_dir = AMRDatabasesManager.get_default_database_directory(
)
epilog = (
"Example:\n"
"\t" + name + " build\n"
"\t\tBuilds a new ResFinder/PointFinder database under " +
self._default_dir + " if it does not exist\n\n" + "\t" + name +
" build --dir databases\n" +
"\t\tBuilds a new ResFinder/PointFinder database under databases/")
arg_parser = self._subparser.add_parser(
'build',
epilog=epilog,
formatter_class=argparse.RawTextHelpFormatter,
help='Downloads and builds databases in the given directory.')
arg_parser.add_argument(
'--dir',
action='store',
dest='destination',
type=str,
help='The directory to download the databases into [' +
self._default_dir + '].',
default=self._default_dir,
required=False)
arg_parser.add_argument(
'--resfinder-commit',
action='store',
dest='resfinder_commit',
type=str,
help='The specific git commit for the resfinder database [latest].',
required=False)
arg_parser.add_argument(
'--pointfinder-commit',
action='store',
dest='pointfinder_commit',
type=str,
help=
'The specific git commit for the pointfinder database [latest].',
required=False)
return arg_parser
def _setup_args(self, arg_parser):
self._default_dir = AMRDatabasesManager.get_default_database_directory(
)
name = self._script_name
epilog = (
"Example:\n"
"\t" + name + " update databases/\n"
"\t\tUpdates the ResFinder/PointFinder database under databases/\n\n"
+ "\t" + name + " update -d\n" +
"\t\tUpdates the default ResFinder/PointFinder database under " +
self._default_dir)
arg_parser = self._subparser.add_parser(
'update',
epilog=epilog,
formatter_class=argparse.RawTextHelpFormatter,
help='Updates databases in the given directories.')
arg_parser.add_argument('-d',
'--update-default',
action='store_true',
dest='update_default',
help='Updates default database directory (' +
self._default_dir + ').',
required=False)
arg_parser.add_argument(
'--resfinder-commit',
action='store',
dest='resfinder_commit',
type=str,
help='The specific git commit for the resfinder database [latest].',
required=False)
arg_parser.add_argument(
'--pointfinder-commit',
action='store',
dest='pointfinder_commit',
type=str,
help=
'The specific git commit for the pointfinder database [latest].',
required=False)
arg_parser.add_argument('directories', nargs='*')
return arg_parser
def run(self, args):
super(Info, self).run(args)
arg_drug_table = ARGDrugTable()
if len(args.directories) == 0:
database_repos = AMRDatabasesManager.create_default_manager(
).get_database_repos()
if not AMRDatabasesManager.is_database_repos_default_commits(
database_repos):
logger.warning(
"Using non-default ResFinder/PointFinder database versions. This may lead to differences in the detected "
+
"AMR genes depending on how the database files are structured."
)
try:
database_info = database_repos.info()
database_info.update(
arg_drug_table.get_resistance_table_info())
sys.stdout.write(get_string_with_spacing(database_info))
except DatabaseNotFoundException as e:
logger.error(
"No database found. Perhaps try restoring the default with 'staramr db restore-default'"
)
else:
for directory in args.directories:
try:
database_repos = AMRDatabasesManager(
directory).get_database_repos()
if not AMRDatabasesManager.is_database_repos_default_commits(
database_repos):
logger.warning(
"Using non-default ResFinder/PointFinder database version [%s]. This may lead to "
+
"differences in the detected AMR genes depending on how the database files are structured.",
directory)
database_info = database_repos.info()
database_info.update(
arg_drug_table.get_resistance_table_info())
sys.stdout.write(get_string_with_spacing(database_info))
except DatabaseNotFoundException as e:
logger.error(
"Database not found in [%s]. Perhaps try building with 'staramr db build --dir %s'",
directory, directory)
class AMRDatabasesManagerIT(unittest.TestCase):
RESFINDER_DEFAULT_COMMIT = 'e8f1eb2585cd9610c4034a54ce7fc4f93aa95535'
POINTFINDER_DEFAULT_COMMIT = '8706a6363bb29e47e0e398c53043b037c24b99a7'
def setUp(self):
self.databases_dir = tempfile.TemporaryDirectory()
self.databases_manager = AMRDatabasesManager(
database_dir=self.databases_dir.name, sub_dirs=True)
def tearDown(self):
self.databases_dir.cleanup()
def testGetHandlerGitStripDir(self):
self.assertIsInstance(self.databases_manager.get_database_handler(),
AMRDatabaseHandlerStripGitDir,
'Invalid instance returned')
def testGetHandlerGit(self):
self.assertIsInstance(
self.databases_manager.get_database_handler(force_use_git=True),
AMRDatabaseHandler, 'Invalid instance returned')
def testSetupDefault(self):
database_handler = self.databases_manager.get_database_handler()
# Verify that databases don't exist beforehand
self.assertFalse(path.exists(database_handler.get_resfinder_dir()),
'resfinder path exists before creation of database')
self.assertFalse(
path.exists(database_handler.get_pointfinder_dir()),
'pointfinder path exists before creation of database')
# Setup default database
self.databases_manager.setup_default()
# Verify that resfinder/pointfinder paths exist
self.assertTrue(path.exists(database_handler.get_resfinder_dir()),
'resfinder path does not exist')
self.assertTrue(path.exists(database_handler.get_resfinder_dir()),
'pointfinder path does not exist')
self.assertTrue(
path.exists(
path.join(database_handler.get_database_dir(), 'info.ini')),
'info file does not exist')
# Verify we've removed the .git directories
self.assertFalse(
path.exists(path.join(database_handler.get_resfinder_dir(),
'.git')),
'resfinder .git directory was not removed')
self.assertFalse(
path.exists(
path.join(database_handler.get_pointfinder_dir(), '.git')),
'pointfinder .git directory was not removed')
config = configparser.ConfigParser()
config.read(path.join(database_handler.get_database_dir(), 'info.ini'))
# Verify that the info.ini file has correct git commits for default database
self.assertEqual(config['GitInfo']['resfinder_db_commit'],
self.RESFINDER_DEFAULT_COMMIT,
'invalid resfinder commit')
self.assertEqual(config['GitInfo']['pointfinder_db_commit'],
self.POINTFINDER_DEFAULT_COMMIT,
'invalid pointfinder commit')
def testRestoreDefault(self):
# Build initial default database
self.databases_manager.setup_default()
# Build updated database
database_handler_git = self.databases_manager.get_database_handler(
force_use_git=True)
database_handler_git.build(
resfinder_commit=self.RESFINDER_DEFAULT_COMMIT,
pointfinder_commit=self.POINTFINDER_DEFAULT_COMMIT)
# Verify that updated database is the one that gets returned by get_database_handler()
database_handler = self.databases_manager.get_database_handler()
self.assertIsInstance(database_handler, AMRDatabaseHandler,
'Invalid instance returned')
self.assertTrue(
path.exists(path.join(database_handler.get_resfinder_dir(),
'.git')),
'Not using git version (updated version) of resfinder database')
self.assertTrue(
path.exists(
path.join(database_handler.get_pointfinder_dir(), '.git')),
'Not using git version (updated version) of pointfinder database')
# Restore default database
self.databases_manager.restore_default()
# Verify that default database (git stripped version) is the one that gets returned by get_database_handler()
database_handler = self.databases_manager.get_database_handler()
self.assertIsInstance(database_handler, AMRDatabaseHandlerStripGitDir,
'Invalid instance returned')
self.assertFalse(
path.exists(path.join(database_handler.get_resfinder_dir(),
'.git')),
'resfinder .git directory was not removed')
self.assertFalse(
path.exists(
path.join(database_handler.get_pointfinder_dir(), '.git')),
'pointfinder .git directory was not removed')
def testIsHandlerDefaultCommitsTrue(self):
# Setup default database
self.databases_manager.setup_default()
database_handler = self.databases_manager.get_database_handler()
self.assertTrue(
AMRDatabasesManager.is_handler_default_commits(database_handler),
"Database is not default")
def testIsHandlerDefaultCommitsFalse(self):
# Setup database
database_handler = self.databases_manager.get_database_handler(
force_use_git=True)
database_handler.update(
resfinder_commit='dc33e2f9ec2c420f99f77c5c33ae3faa79c999f2')
self.assertFalse(
AMRDatabasesManager.is_handler_default_commits(database_handler),
"Database is default")
class AMRDatabasesManagerIT(unittest.TestCase):
RESFINDER_DEFAULT_COMMIT = 'e8f1eb2585cd9610c4034a54ce7fc4f93aa95535'
POINTFINDER_DEFAULT_COMMIT = '8706a6363bb29e47e0e398c53043b037c24b99a7'
PLASMIDFINDER_DEFAULT_COMMIT = '81919954cbedaff39056610ab584ab4c06011ed8'
def setUp(self):
self.databases_dir = tempfile.TemporaryDirectory()
self.databases_manager = AMRDatabasesManager(
database_dir=self.databases_dir.name, sub_dirs=True)
def tearDown(self):
self.databases_dir.cleanup()
def testSetupDefault(self):
blast_database_repos = self.databases_manager.get_database_repos()
# Verify that databases don't exist beforehand
self.assertFalse(
path.exists(blast_database_repos.get_repo_dir('resfinder')),
'resfinder path exists before creation of database')
self.assertFalse(
path.exists(blast_database_repos.get_repo_dir('pointfinder')),
'pointfinder path exists before creation of database')
self.assertFalse(
path.exists(blast_database_repos.get_repo_dir('plasmidfinder')),
'plasmidfinder path exists before creation of database')
# Setup default database
self.databases_manager.setup_default()
# Verify that resfinder/pointfinder paths exist
self.assertTrue(
path.exists(blast_database_repos.get_repo_dir('resfinder')),
'resfinder path does not exist')
self.assertTrue(
path.exists(blast_database_repos.get_repo_dir('pointfinder')),
'pointfinder path does not exist')
self.assertTrue(
path.exists(blast_database_repos.get_repo_dir('plasmidfinder')),
'plasmidfinder path does not exist')
self.assertTrue(
path.exists(
path.join(blast_database_repos.get_database_dir(),
'resfinder-info.ini')),
'resfinder info file does not exist')
self.assertTrue(
path.exists(
path.join(blast_database_repos.get_database_dir(),
'pointfinder-info.ini')),
'pointfinder info file does not exist')
self.assertTrue(
path.exists(
path.join(blast_database_repos.get_database_dir(),
'plasmidfinder-info.ini')),
'plasmidfinder info file does not exist')
# Verify we've removed the .git directories
self.assertFalse(
path.exists(
path.join(blast_database_repos.get_repo_dir('resfinder'),
'.git')), 'resfinder .git directory was not removed')
self.assertFalse(
path.exists(
path.join(blast_database_repos.get_repo_dir('pointfinder'),
'.git')),
'pointfinder .git directory was not removed')
self.assertFalse(
path.exists(
path.join(blast_database_repos.get_repo_dir('plasmidfinder'),
'.git')),
'plasmidfinder .git directory was not removed')
config = configparser.ConfigParser()
# Verify that the info.ini file has correct git commits for default database
config.read(
path.join(blast_database_repos.get_database_dir(),
'resfinder-info.ini'))
self.assertEqual(config['GitInfo']['resfinder_db_commit'],
self.RESFINDER_DEFAULT_COMMIT,
'invalid resfinder commit')
config.read(
path.join(blast_database_repos.get_database_dir(),
'pointfinder-info.ini'))
self.assertEqual(config['GitInfo']['pointfinder_db_commit'],
self.POINTFINDER_DEFAULT_COMMIT,
'invalid pointfinder commit')
config.read(
path.join(blast_database_repos.get_database_dir(),
'plasmidfinder-info.ini'))
self.assertEqual(config['GitInfo']['plasmidfinder_db_commit'],
self.PLASMIDFINDER_DEFAULT_COMMIT,
'invalid plasmidfinder commit')
def testRestoreDefault(self):
# Build initial default database
self.databases_manager.setup_default()
# Build updated database
blast_database_repos_git = self.databases_manager.get_database_repos(
force_use_git=True)
blast_database_repos_git.build({
'resfinder':
self.RESFINDER_DEFAULT_COMMIT,
'pointfinder':
self.POINTFINDER_DEFAULT_COMMIT,
'plasmidfinder':
self.PLASMIDFINDER_DEFAULT_COMMIT
})
# Verify that updated database is the one that gets returned by get_database_handler()
blast_database_repos = self.databases_manager.get_database_repos()
self.assertFalse(blast_database_repos.is_dist(), 'Invalid is_dist')
self.assertEqual(blast_database_repos.get_database_dir(),
path.join(self.databases_dir.name, 'update'),
'Invalid database directory')
self.assertTrue(
path.exists(
path.join(blast_database_repos.get_repo_dir('resfinder'),
'.git')),
'Not using git version (updated version) of resfinder database')
self.assertTrue(
path.exists(
path.join(blast_database_repos.get_repo_dir('pointfinder'),
'.git')),
'Not using git version (updated version) of pointfinder database')
self.assertTrue(
path.exists(
path.join(blast_database_repos.get_repo_dir('plasmidfinder'),
'.git')),
'Not using git version (updated version) of plasmidfinder database'
)
# Restore default database
self.databases_manager.restore_default()
# Verify that default database (git stripped version) is the one that gets returned by get_database_handler()
blast_database_repos = self.databases_manager.get_database_repos()
self.assertTrue(blast_database_repos.is_dist(), 'Invalid is_dist')
self.assertEqual(blast_database_repos.get_database_dir(),
path.join(self.databases_dir.name, 'dist'),
'Invalid database directory')
self.assertFalse(
path.exists(
path.join(blast_database_repos.get_repo_dir('resfinder'),
'.git')), 'resfinder .git directory was not removed')
self.assertFalse(
path.exists(
path.join(blast_database_repos.get_repo_dir('pointfinder'),
'.git')),
'pointfinder .git directory was not removed')
self.assertFalse(
path.exists(
path.join(blast_database_repos.get_repo_dir('plasmidfinder'),
'.git')),
'plasmidfinder .git directory was not removed')
def testIsHandlerDefaultCommitsTrue(self):
# Setup default database
self.databases_manager.setup_default()
blast_database_repos = self.databases_manager.get_database_repos()
self.assertTrue(
AMRDatabasesManager.is_database_repos_default_commits(
blast_database_repos), "Database is not default")
def testIsHandlerDefaultCommitsFalse(self):
# Setup database
blast_database_repos = self.databases_manager.get_database_repos(
force_use_git=True)
blast_database_repos.update(
{'resfinder': 'dc33e2f9ec2c420f99f77c5c33ae3faa79c999f2'})
self.assertFalse(
AMRDatabasesManager.is_database_repos_default_commits(
blast_database_repos), "Database is default")
def _setup_args(self, arg_parser):
name = self._script_name
epilog = (
"Example:\n"
"\t" + name + " search -o out *.fasta\n"
"\t\tSearches the files *.fasta for AMR genes using only the ResFinder database, storing results in the out/ directory.\n\n"
+ "\t" + name +
" search --pointfinder-organism salmonella --output-excel results.xlsx *.fasta\n"
+
"\t\tSearches *.fasta for AMR genes using ResFinder and PointFinder database with the passed organism, storing results in results.xlsx."
)
arg_parser = self._subparser.add_parser(
'search',
epilog=epilog,
formatter_class=argparse.RawTextHelpFormatter,
help='Search for AMR genes')
self._default_database_dir = AMRDatabasesManager.get_default_database_directory(
)
cpu_count = multiprocessing.cpu_count()
arg_parser.add_argument(
'--pointfinder-organism',
action='store',
dest='pointfinder_organism',
type=str,
help='The organism to use for pointfinder {' +
', '.join(PointfinderBlastDatabase.get_available_organisms()) +
'}. Defaults to disabling search for point mutations. [None].',
default=None,
required=False)
arg_parser.add_argument(
'-d',
'--database',
action='store',
dest='database',
type=str,
help=
'The directory containing the resfinder/pointfinder databases [' +
self._default_database_dir + '].',
default=self._default_database_dir,
required=False)
arg_parser.add_argument(
'-n',
'--nprocs',
action='store',
dest='nprocs',
type=int,
help='The number of processing cores to use [' + str(cpu_count) +
'].',
default=cpu_count,
required=False)
threshold_group = arg_parser.add_argument_group('BLAST Thresholds')
threshold_group.add_argument(
'--pid-threshold',
action='store',
dest='pid_threshold',
type=float,
help='The percent identity threshold [98.0].',
default=98.0,
required=False)
threshold_group.add_argument(
'--percent-length-overlap-resfinder',
action='store',
dest='plength_threshold_resfinder',
type=float,
help='The percent length overlap for resfinder results [60.0].',
default=60.0,
required=False)
threshold_group.add_argument(
'--percent-length-overlap-pointfinder',
action='store',
dest='plength_threshold_pointfinder',
type=float,
help='The percent length overlap for pointfinder results [95.0].',
default=95.0,
required=False)
report_group = arg_parser.add_argument_group('Reporting options')
report_group.add_argument(
'--no-exclude-genes',
action='store_true',
dest='no_exclude_genes',
help=
'Disable the default exclusion of some genes from ResFinder/PointFinder [False].',
required=False)
report_group.add_argument(
'--exclude-genes-file',
action='store',
dest='exclude_genes_file',
help=
'A containing a list of ResFinder/PointFinder gene names to exclude from results [{}].'
.format(ExcludeGenesList.get_default_exclude_file()),
default=ExcludeGenesList.get_default_exclude_file(),
required=False)
report_group.add_argument(
'--exclude-negatives',
action='store_true',
dest='exclude_negatives',
help=
'Exclude negative results (those sensitive to antimicrobials) [False].',
required=False)
report_group.add_argument(
'--exclude-resistance-phenotypes',
action='store_true',
dest='exclude_resistance_phenotypes',
help='Exclude predicted antimicrobial resistances [False].',
required=False)
report_group.add_argument(
'--report-all-blast',
action='store_true',
dest='report_all_blast',
help='Report all blast hits (vs. only top blast hits) [False].',
required=False)
output_group = arg_parser.add_argument_group(
title='Output',
description=
'Use either --output-dir or specify individual output files')
output_group.add_argument(
'-o',
'--output-dir',
action='store',
dest='output_dir',
type=str,
help="The output directory for results [None].",
default=None,
required=False)
output_group.add_argument(
'--output-summary',
action='store',
dest='output_summary',
type=str,
help=
"The name of the output file containing the summary results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-resfinder',
action='store',
dest='output_resfinder',
type=str,
help=
"The name of the output file containing the resfinder results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-pointfinder',
action='store',
dest='output_pointfinder',
type=str,
help=
"The name of the output file containing the pointfinder results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-settings',
action='store',
dest='output_settings',
type=str,
help=
"The name of the output file containing the settings. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-excel',
action='store',
dest='output_excel',
type=str,
help=
"The name of the output file containing the excel results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-hits-dir',
action='store',
dest='hits_output_dir',
type=str,
help=
"The name of the directory to contain the BLAST hit files. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
arg_parser.add_argument('files', nargs='+')
return arg_parser
def _setup_args(self, arg_parser):
name = self._script_name
epilog = (
"Example:\n"
"\t" + name + " search -o out *.fasta\n"
"\t\tSearches the files *.fasta for AMR genes using only the ResFinder database, storing results in the out/ directory.\n\n"
+ "\t" + name +
" search --pointfinder-organism salmonella --output-excel results.xlsx *.fasta\n"
+
"\t\tSearches *.fasta for AMR genes using ResFinder and PointFinder database with the passed organism, storing results in results.xlsx."
)
arg_parser = self._subparser.add_parser(
'search',
epilog=epilog,
formatter_class=argparse.RawTextHelpFormatter,
help='Search for AMR genes')
self._default_database_dir = AMRDatabasesManager.get_default_database_directory(
)
cpu_count = multiprocessing.cpu_count()
arg_parser.add_argument(
'--pointfinder-organism',
action='store',
dest='pointfinder_organism',
type=str,
help='The organism to use for pointfinder {' +
', '.join(PointfinderBlastDatabase.get_available_organisms()) +
'}. Defaults to disabling search for point mutations. [None].',
default=None,
required=False)
arg_parser.add_argument(
'--plasmidfinder-database-type',
action='store',
dest='plasmidfinder_database_type',
type=str,
help='The database type to use for plasmidfinder {' +
', '.join(PlasmidfinderBlastDatabase.get_available_databases()) +
'}. Defaults to using all available database types to search for plasmids. [None].',
default=None,
required=False)
arg_parser.add_argument(
'-d',
'--database',
action='store',
dest='database',
type=str,
help=
'The directory containing the resfinder/pointfinder/plasmidfinder databases ['
+ self._default_database_dir + '].',
default=self._default_database_dir,
required=False)
arg_parser.add_argument(
'-n',
'--nprocs',
action='store',
dest='nprocs',
type=int,
help='The number of processing cores to use [' + str(cpu_count) +
'].',
default=cpu_count,
required=False)
arg_parser.add_argument('--ignore-invalid-files',
action='store_true',
dest='ignore_valid_files',
help='Skips over invalid input files',
required=False)
arg_parser.add_argument(
'--mlst-scheme',
action='store',
dest='mlst_scheme',
help=
'Specify scheme name, visit https://github.com/tseemann/mlst/blob/master/db/scheme_species_map.tab for supported scheme genus available. [None] ',
required=False)
arg_parser.add_argument(
'--genome-size-lower-bound',
action='store',
dest='genome_size_lower_bound',
type=int,
help=
'The lower bound for our genome size for the quality metrics. Defaults to 4 Mbp. [4000000].',
default=4000000,
required=False)
arg_parser.add_argument(
'--genome-size-upper-bound',
action='store',
dest='genome_size_upper_bound',
type=int,
help=
'The upper bound for our genome size for the quality metrics. Defaults to 6 Mbp. [6000000].',
default=6000000,
required=False)
arg_parser.add_argument(
'--minimum-N50-value',
action='store',
dest='minimum_N50_value',
type=int,
help=
'The minimum N50 value for the quality metrics. Defaults to 10000. [10000].',
default=10000,
required=False)
arg_parser.add_argument(
'--minimum-contig-length',
action='store',
dest='minimum_contig_length',
type=int,
help=
'The minimum contig length for the quality metrics. Defaults to 300 bp. [300].',
default=300,
required=False)
arg_parser.add_argument(
'--unacceptable-number-contigs',
action='store',
dest='unacceptable_num_contigs',
type=int,
help=
'The minimum, unacceptable number of contigs which are equal to or above the minimum contig length for our quality metrics. Defaults to 1000. [1000].',
default=1000,
required=False)
threshold_group = arg_parser.add_argument_group('BLAST Thresholds')
threshold_group.add_argument(
'--pid-threshold',
action='store',
dest='pid_threshold',
type=float,
help='The percent identity threshold [98.0].',
default=98.0,
required=False)
threshold_group.add_argument(
'--percent-length-overlap-resfinder',
action='store',
dest='plength_threshold_resfinder',
type=float,
help='The percent length overlap for resfinder results [60.0].',
default=60.0,
required=False)
threshold_group.add_argument(
'--percent-length-overlap-pointfinder',
action='store',
dest='plength_threshold_pointfinder',
type=float,
help='The percent length overlap for pointfinder results [95.0].',
default=95.0,
required=False)
threshold_group.add_argument(
'--percent-length-overlap-plasmidfinder',
action='store',
dest='plength_threshold_plasmidfinder',
type=float,
help='The percent length overlap for resfinder results [60.0].',
default=60.0,
required=False)
report_group = arg_parser.add_argument_group('Reporting options')
report_group.add_argument(
'--no-exclude-genes',
action='store_true',
dest='no_exclude_genes',
help=
'Disable the default exclusion of some genes from ResFinder/PointFinder/PlasmidFinder [False].',
required=False)
report_group.add_argument(
'--exclude-genes-file',
action='store',
dest='exclude_genes_file',
help=
'A containing a list of ResFinder/PointFinder/PlasmidFinder gene names to exclude from results [{}].'
.format(ExcludeGenesList.get_default_exclude_file()),
default=ExcludeGenesList.get_default_exclude_file(),
required=False)
report_group.add_argument(
'--exclude-negatives',
action='store_true',
dest='exclude_negatives',
help=
'Exclude negative results (those sensitive to antimicrobials) [False].',
required=False)
report_group.add_argument(
'--exclude-resistance-phenotypes',
action='store_true',
dest='exclude_resistance_phenotypes',
help='Exclude predicted antimicrobial resistances [False].',
required=False)
report_group.add_argument(
'--report-all-blast',
action='store_true',
dest='report_all_blast',
help='Report all blast hits (vs. only top blast hits) [False].',
required=False)
output_group = arg_parser.add_argument_group(
title='Output',
description=
'Use either --output-dir or specify individual output files')
output_group.add_argument(
'-o',
'--output-dir',
action='store',
dest='output_dir',
type=str,
help="The output directory for results [None].",
default=None,
required=False)
output_group.add_argument(
'--output-summary',
action='store',
dest='output_summary',
type=str,
help=
"The name of the output file containing the summary results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-detailed-summary',
action='store',
dest='output_detailed_summary',
type=str,
help=
"The name of the output file containing the detailed summary results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-resfinder',
action='store',
dest='output_resfinder',
type=str,
help=
"The name of the output file containing the resfinder results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-pointfinder',
action='store',
dest='output_pointfinder',
type=str,
help=
"The name of the output file containing the pointfinder results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-plasmidfinder',
action='store',
dest='output_plasmidfinder',
type=str,
help=
"The name of the output file containing the plasmidfinder results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-settings',
action='store',
dest='output_settings',
type=str,
help=
"The name of the output file containing the settings. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-excel',
action='store',
dest='output_excel',
type=str,
help=
"The name of the output file containing the excel results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-hits-dir',
action='store',
dest='hits_output_dir',
type=str,
help=
"The name of the directory to contain the BLAST hit files. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
output_group.add_argument(
'--output-mlst',
action='store',
dest='output_mlst',
type=str,
help=
"The name of the output file containing the mlst results. Not be be used with '--output-dir'. [None]",
default=None,
required=False)
arg_parser.add_argument('files', nargs='+')
return arg_parser
def setUp(self):
self.databases_dir = tempfile.TemporaryDirectory()
self.databases_manager = AMRDatabasesManager(
database_dir=self.databases_dir.name, sub_dirs=True)
def run(self, args):
super(Search, self).run(args)
if (len(args.files) == 0):
raise CommandParseException("Must pass a fasta file to process",
self._root_arg_parser,
print_help=True)
for file in args.files:
if not path.exists(file):
raise CommandParseException(
'File [' + file + '] does not exist',
self._root_arg_parser)
if not path.isdir(args.database):
if args.database == self._default_database_dir:
raise CommandParseException(
"Default database does not exist. Perhaps try restoring with 'staramr db restore-default'",
self._root_arg_parser)
else:
raise CommandParseException(
"Database directory [" + args.database +
"] does not exist. Perhaps try building with" +
"'staramr db build --dir " + args.database + "'",
self._root_arg_parser)
if args.database == AMRDatabasesManager.get_default_database_directory(
):
database_repos = AMRDatabasesManager.create_default_manager(
).get_database_repos()
else:
database_repos = AMRDatabasesManager(
args.database).get_database_repos()
if not AMRDatabasesManager.is_database_repos_default_commits(
database_repos):
logger.warning(
"Using non-default ResFinder/PointFinder. This may lead to differences in the detected "
+
"AMR genes depending on how the database files are structured."
)
resfinder_database = database_repos.build_blast_database('resfinder')
if (args.pointfinder_organism):
if args.pointfinder_organism not in PointfinderBlastDatabase.get_available_organisms(
):
raise CommandParseException(
"The only Pointfinder organism(s) currently supported are "
+ str(PointfinderBlastDatabase.get_available_organisms()),
self._root_arg_parser)
pointfinder_database = database_repos.build_blast_database(
'pointfinder', {'organism': args.pointfinder_organism})
else:
logger.info(
"No --pointfinder-organism specified. Will not search the PointFinder databases"
)
pointfinder_database = None
if (args.plasmidfinder_database_type):
if args.plasmidfinder_database_type not in PlasmidfinderBlastDatabase.get_available_databases(
):
raise CommandParseException(
"The only Plasmidfinder databases that are currently supported are "
+
str(PlasmidfinderBlastDatabase.get_available_databases()),
self._root_arg_parser)
plasmidfinder_database = database_repos.build_blast_database(
'plasmidfinder',
{'database_type': args.plasmidfinder_database_type})
else:
logger.info(
"No --plasmidfinder-database-type specified. Will search the entire PlasmidFinder database"
)
plasmidfinder_database = database_repos.build_blast_database(
'plasmidfinder')
hits_output_dir = None
output_summary = None
output_detailed_summary = None
output_resfinder = None
output_pointfinder = None
output_plasmidfinder = None
output_mlst = None
output_excel = None
output_settings = None
if args.output_dir:
if path.exists(args.output_dir):
raise CommandParseException(
"Output directory [" + args.output_dir +
"] already exists", self._root_arg_parser)
elif args.output_summary or args.output_detailed_summary or args.output_resfinder or args.output_pointfinder or args.output_plasmidfinder or args.output_excel or \
args.hits_output_dir:
raise CommandParseException(
'You cannot use --output-[type] with --output-dir',
self._root_arg_parser)
else:
mkdir(args.output_dir)
hits_output_dir = path.join(args.output_dir, 'hits')
output_resfinder = path.join(args.output_dir, "resfinder.tsv")
output_pointfinder = path.join(args.output_dir,
"pointfinder.tsv")
output_plasmidfinder = path.join(args.output_dir,
"plasmidfinder.tsv")
output_summary = path.join(args.output_dir, "summary.tsv")
output_detailed_summary = path.join(args.output_dir,
"detailed_summary.tsv")
output_mlst = path.join(args.output_dir, "mlst.tsv")
output_settings = path.join(args.output_dir, "settings.txt")
output_excel = path.join(args.output_dir, 'results.xlsx')
mkdir(hits_output_dir)
logger.info(
"--output-dir set. All files will be output to [%s]",
args.output_dir)
elif args.output_summary or args.output_excel or args.output_detailed_summary:
logger.info(
'--output-dir not set. Files will be output to the respective --output-[type] setting'
)
output_resfinder = args.output_resfinder
output_pointfinder = args.output_pointfinder
output_plasmidfinder = args.output_plasmidfinder
output_summary = args.output_summary
output_detailed_summary = args.output_detailed_summary
output_mlst = args.output_mlst
output_settings = args.output_settings
output_excel = args.output_excel
hits_output_dir = args.hits_output_dir
if hits_output_dir is not None:
if path.exists(
hits_output_dir) and not path.isdir(hits_output_dir):
raise CommandParseException(
"--output-hits-dir [" + hits_output_dir +
"] exists and is not a directory",
self._root_arg_parser)
elif path.exists(hits_output_dir):
logger.debug(
"Found --output-hits-dir [%s] and is a directory. Will write hits here",
hits_output_dir)
else:
logger.debug("Making directory [%s]", hits_output_dir)
mkdir(hits_output_dir)
else:
raise CommandParseException(
'You must set one of --output-dir, --output-summary, --output-detailed-summary, or --output-excel',
self._root_arg_parser)
if args.no_exclude_genes:
logger.info(
"--no-exclude-genes enabled. Will not exclude any ResFinder/PointFinder genes."
)
exclude_genes = []
else:
if not path.exists(args.exclude_genes_file):
raise CommandParseException(
'--exclude-genes-file [{}] does not exist'.format(
args.exclude_genes_file), self._root_arg_parser)
else:
logger.info(
"Will exclude ResFinder/PointFinder genes listed in [%s]. Use --no-exclude-genes to disable",
args.exclude_genes_file)
exclude_genes = ExcludeGenesList(
args.exclude_genes_file).tolist()
results = self._generate_results(
database_repos=database_repos,
resfinder_database=resfinder_database,
pointfinder_database=pointfinder_database,
plasmidfinder_database=plasmidfinder_database,
nprocs=args.nprocs,
include_negatives=not args.exclude_negatives,
include_resistances=not args.exclude_resistance_phenotypes,
hits_output=hits_output_dir,
pid_threshold=args.pid_threshold,
plength_threshold_resfinder=args.plength_threshold_resfinder,
plength_threshold_pointfinder=args.plength_threshold_pointfinder,
plength_threshold_plasmidfinder=args.
plength_threshold_plasmidfinder,
report_all_blast=args.report_all_blast,
genes_to_exclude=exclude_genes,
files=args.files,
ignore_invalid_files=args.ignore_valid_files,
mlst_scheme=args.mlst_scheme,
genome_size_lower_bound=args.genome_size_lower_bound,
genome_size_upper_bound=args.genome_size_upper_bound,
minimum_N50_value=args.minimum_N50_value,
minimum_contig_length=args.minimum_contig_length,
unacceptable_num_contigs=args.unacceptable_num_contigs)
amr_detection = results['results']
settings = results['settings']
if output_resfinder:
logger.info("Writing resfinder to [%s]", output_resfinder)
with open(output_resfinder, 'w') as fh:
self._print_dataframe_to_text_file_handle(
amr_detection.get_resfinder_results(), fh)
else:
logger.info(
"--output-dir or --output-resfinder unset. No resfinder file will be written"
)
if args.pointfinder_organism and output_pointfinder:
logger.info("Writing pointfinder to [%s]", output_pointfinder)
with open(output_pointfinder, 'w') as fh:
self._print_dataframe_to_text_file_handle(
amr_detection.get_pointfinder_results(), fh)
else:
logger.info(
"--output-dir or --output-pointfinder unset. No pointfinder file will be written"
)
if output_plasmidfinder:
logger.info("Writing plasmidfinder to [%s]", output_plasmidfinder)
with open(output_plasmidfinder, 'w') as fh:
self._print_dataframe_to_text_file_handle(
amr_detection.get_plasmidfinder_results(), fh)
else:
logger.info(
"--output-dir or --output-plasmidfinder unset. No plasmidfinder file will be written"
)
if output_summary:
logger.info("Writing summary to [%s]", output_summary)
with open(output_summary, 'w') as fh:
self._print_dataframe_to_text_file_handle(
amr_detection.get_summary_results(), fh)
else:
logger.info(
"--output-dir or --output-summary unset. No summary file will be written"
)
if output_mlst:
logger.info("Writing MLST summary to [%s]", output_mlst)
with open(output_mlst, 'w') as fh:
self._print_dataframe_to_text_file_handle(
amr_detection.get_mlst_results(), fh)
else:
logger.info(
"--output-dir or --output-mlst unset. No mlst file will be written"
)
if output_detailed_summary:
logger.info("Writing detailed summary to [%s]",
output_detailed_summary)
with open(output_detailed_summary, 'w') as fh:
self._print_dataframe_to_text_file_handle(
amr_detection.get_detailed_summary_results(), fh)
else:
logger.info(
"--output-dir or --output-detailed-summary unset. No detailed summary file will be written"
)
if output_settings:
logger.info("Writing settings to [%s]", output_settings)
self._print_settings_to_file(settings, output_settings)
else:
logger.info(
"--output-dir or --output-settings unset. No settings file will be written"
)
if output_excel:
logger.info("Writing Excel to [%s]", output_excel)
settings_dataframe = pd.DataFrame.from_dict(settings,
orient='index')
settings_dataframe.index.name = 'Key'
settings_dataframe.set_axis(['Value'],
axis='columns',
inplace=True)
self._print_dataframes_to_excel(
output_excel, amr_detection.get_summary_results(),
amr_detection.get_resfinder_results(),
amr_detection.get_pointfinder_results(),
amr_detection.get_plasmidfinder_results(),
amr_detection.get_detailed_summary_results(),
amr_detection.get_mlst_results(), settings_dataframe,
args.minimum_contig_length)
else:
logger.info(
"--output-dir or --output-excel unset. No excel file will be written"
)
if hits_output_dir:
logger.info("BLAST hits are stored in [%s]", hits_output_dir)
else:
logger.info(
"--output-dir or --output-hits-dir not set. No BLAST hits will be saved."
)
