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 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 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)
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."
)