def main(args=None, log_level=None):
"""
main entry point to integron_split
:param str args: the arguments passed on the command line
:param log_level: the output verbosity
:type log_level: a positive int or a string among 'DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL'
"""
global _log
args = sys.argv[1:] if args is None else args
parsed_args = parse_args(args)
integron_finder.init_logger(log_file=os.path.join(parsed_args.outdir, 'integron_split.out'),
out=not parsed_args.mute)
_log = colorlog.getLogger('integron_finder.split')
if not log_level:
# logs are specify from args options
logger_set_level(utils.log_level(parsed_args.verbose, parsed_args.quiet))
else:
# used by unit tests to mute or unmute logs
logger_set_level(log_level)
chunk_names = split(parsed_args.replicon, chunk=parsed_args.chunk, outdir=parsed_args.outdir)
print(' '.join(chunk_names))
def main(args=None, log_level=None):
"""
main entry point to integron_split
:param str args: the arguments passed on the command line
:param log_level: the output verbosity
:type log_level: a positive int or a string among 'DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL'
"""
global _log
args = sys.argv[1:] if args is None else args
parsed_args = parse_args(args)
integron_finder.init_logger(log_file=os.path.join(parsed_args.outdir,
'integron_split.out'),
out=not parsed_args.mute)
_log = colorlog.getLogger('integron_finder.split')
if not log_level:
# logs are specify from args options
logger_set_level(
utils.log_level(parsed_args.verbose, parsed_args.quiet))
else:
# used by unit tests to mute or unmute logs
logger_set_level(log_level)
chunk_names = split(parsed_args.replicon,
chunk=parsed_args.chunk,
outdir=parsed_args.outdir)
print(' '.join(chunk_names))
def setUp(self):
logger_set_level('INFO')
self.tmp_dir = os.path.join(tempfile.gettempdir(),
'tmp_test_integron_finder')
if os.path.isdir(self.tmp_dir):
shutil.rmtree(self.tmp_dir)
os.makedirs(self.tmp_dir)
def test_merge_no_files(self):
outfile = os.path.join(self.out_dir, 'merged.integrons')
for rep, res_dir in zip(self.replicons, self.res_dirs):
os.unlink(os.path.join(res_dir, "{}.integrons".format(rep)))
exp_msg = 'No integrons file to merge'
with self.assertRaises(IntegronError) as ctx:
# 100 is to disable CRITICAL log message
logger_set_level(100)
merge.merge_integrons(outfile, *self.res_dirs)
self.assertEqual(str(ctx.exception), exp_msg)
def test_merge_no_files(self):
outfile = os.path.join(self.out_dir, 'merged.integrons')
for rep, res_dir in zip(self.replicons, self.res_dirs):
os.unlink(os.path.join(res_dir, "{}.integrons".format(rep)))
exp_msg = 'No integrons file to merge'
with self.assertRaises(IntegronError) as ctx:
# 100 is to disable CRITICAL log message
logger_set_level(100)
merge.merge_integrons(outfile, *self.res_dirs)
self.assertEqual(str(ctx.exception), exp_msg)
def set_log_level(cls, level):
levels = {'NOTSET': colorlog.logging.logging.NOTSET,
'DEBUG': colorlog.logging.logging.DEBUG,
'INFO': colorlog.logging.logging.INFO,
'WARNING': colorlog.logging.logging.WARNING,
'ERROR': colorlog.logging.logging.ERROR,
'CRITICAL': colorlog.logging.logging.CRITICAL,
}
if level in levels:
level = levels[level]
elif not isinstance(level, int):
raise IntegronError("Level must be {} or a positive integer")
elif level < 0:
raise IntegronError("Level must be {} or a positive integer")
logger_set_level(level)
def set_log_level(cls, level):
levels = {
'NOTSET': colorlog.logging.logging.NOTSET,
'DEBUG': colorlog.logging.logging.DEBUG,
'INFO': colorlog.logging.logging.INFO,
'WARNING': colorlog.logging.logging.WARNING,
'ERROR': colorlog.logging.logging.ERROR,
'CRITICAL': colorlog.logging.logging.CRITICAL,
}
if level in levels:
level = levels[level]
elif not isinstance(level, int):
raise IntegronError("Level must be {} or a positive integer")
elif level < 0:
raise IntegronError("Level must be {} or a positive integer")
logger_set_level(level)
def main(args=None, log_level=None):
"""
main entry point to integron_merge
:param str args: the arguments passed on the command line
:param log_level: the output verbosity
:type log_level: a positive int or a string among 'DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL'
"""
global _log
args = sys.argv[1:] if args is None else args
parsed_args = parse_args(args)
integron_finder.init_logger()
_log = colorlog.getLogger('integron_finder.merge')
if not log_level:
# logs are specify from args options
if not log_level:
# logs are specify from args options
logger_set_level(
utils.log_level(parsed_args.verbose, parsed_args.quiet))
else:
# used by unit tests to mute or unmute logs
logger_set_level(log_level)
outdir = os.path.realpath(parsed_args.outdir)
if os.path.exists(outdir):
if not os.path.isdir(outdir):
msg = "'{}' already exists and is not a directory".format(outdir)
_log.critical(msg)
raise IOError(msg)
else:
os.makedirs(parsed_args.outdir)
integron_file_out = os.path.join(outdir,
parsed_args.outfile + ".integrons")
merge_integrons(integron_file_out, *parsed_args.results)
summary_file_out = os.path.join(outdir, parsed_args.outfile + ".summary")
merge_summary(summary_file_out, *parsed_args.results)
copy_file(outdir, '.gbk', *parsed_args.results)
copy_file(outdir, '.pdf', *parsed_args.results)
copy_dir(outdir, 'tmp_*', *parsed_args.results)
def main(args=None, log_level=None):
"""
main entry point to integron_merge
:param str args: the arguments passed on the command line
:param log_level: the output verbosity
:type log_level: a positive int or a string among 'DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL'
"""
global _log
args = sys.argv[1:] if args is None else args
parsed_args = parse_args(args)
integron_finder.init_logger()
_log = colorlog.getLogger('integron_finder.merge')
if not log_level:
# logs are specify from args options
if not log_level:
# logs are specify from args options
logger_set_level(utils.log_level(parsed_args.verbose, parsed_args.quiet))
else:
# used by unit tests to mute or unmute logs
logger_set_level(log_level)
outdir = os.path.realpath(parsed_args.outdir)
if os.path.exists(outdir):
if not os.path.isdir(outdir):
msg = "'{}' already exists and is not a directory".format(outdir)
_log.critical(msg)
raise IOError(msg)
else:
os.makedirs(parsed_args.outdir)
integron_file_out = os.path.join(outdir, parsed_args.outfile + ".integrons")
merge_integrons(integron_file_out, *parsed_args.results)
summary_file_out = os.path.join(outdir, parsed_args.outfile + ".summary")
merge_summary(summary_file_out, *parsed_args.results)
copy_file(outdir, '.gbk', *parsed_args.results)
copy_file(outdir, '.pdf', *parsed_args.results)
copy_dir(outdir, 'tmp_*', *parsed_args.results)
def main(args=None, loglevel=None):
"""
main entry point to integron_finder
:param str args: the arguments passed on the command line
:param loglevel: the output verbosity
:type loglevel: a positive int or a string among 'DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL'
"""
global _log
args = sys.argv[1:] if args is None else args
config = parse_args(args)
###################################
# Prepare directories for results #
###################################
# need to create directory before to init logger
# as we write log in integron_finder.out in this dir
if not os.path.exists(config.outdir):
os.mkdir(config.outdir)
else:
if not os.path.isdir(config.outdir):
msg = "outdir '{}' already exists and is not a directory".format(config.outdir)
# _log.critical(msg)
# we can not log it because logger are not initialized yet.
raise IsADirectoryError(msg)
if not os.path.exists(config.result_dir):
os.mkdir(config.result_dir)
else:
if not os.path.isdir(config.result_dir):
msg = "result dir '{}' already exists and is not a directory".format(config.outdir)
# _log.critical(msg)
# we can not log it because logger are not initialized yet.
raise IsADirectoryError(msg)
elif not os.access(config.result_dir, os.W_OK):
msg = "result dir '{}' already exists and is not writable".format(config.outdir)
# _log.critical(msg)
# we can not log it because logger are not initialized yet.
raise PermissionError(msg)
####################
# init the loggers #
####################
log_file = os.path.join(config.result_dir, 'integron_finder.out')
integron_finder.init_logger(log_file=log_file,
out=not config.mute)
_log = colorlog.getLogger('integron_finder')
if not loglevel:
# logs are specify from args options
logger_set_level(config.log_level)
else:
# used by unit tests to mute or unmute logs
logger_set_level(loglevel)
#######################################
# do last config check before running #
#######################################
if config.cmsearch is None:
msg = """cannot find 'cmsearch' in PATH.
Please install infernal package or setup 'cmsearch' binary path with --cmsearch option"""
_log.critical(msg)
raise RuntimeError(msg)
if config.hmmsearch is None:
msg = """cannot find 'hmmsearch' in PATH.
Please install hmmer package or setup 'hmmsearch' binary path with --hmmsearch option"""
_log.critical(msg)
raise RuntimeError(msg)
if config.prodigal is None:
msg = """cannot find 'prodigal' in PATH.
Please install prodigal package or setup 'prodigal' binary path with --prodigal option"""
_log.critical(msg)
raise RuntimeError(msg)
################
# print Header #
################
log_header = colorlog.getLogger('integron_finder.header')
logging = colorlog.logging.logging
handlers = []
header_log_file = logging.FileHandler(log_file)
handlers.append(header_log_file)
if not config.mute:
header_stream = colorlog.StreamHandler(sys.stdout)
handlers.append(header_stream)
formatter = colorlog.ColoredFormatter("%(message)s")
for h in handlers:
h.setFormatter(formatter)
log_header.addHandler(h)
log_header.setLevel(colorlog.logging.logging.INFO)
log_header.propagate = False
log_header.info(header(args))
with utils.FastaIterator(config.input_seq_path, dist_threshold=config.distance_threshold) as sequences_db:
################
# set topology #
################
default_topology = 'circ' if len(sequences_db) == 1 else 'lin'
if config.linear:
default_topology = 'lin'
elif config.circular:
default_topology = 'circ'
# the both options are mutually exclusive
topologies = Topology(default_topology, topology_file=config.topology_file)
# allow sequences_db to inject topology information
# in seq.topology attribute
sequences_db.topologies = topologies
##############
# do the job #
##############
sequences_db_len = len(sequences_db)
all_integrons = []
all_summaries = []
for rep_no, replicon in enumerate(sequences_db, 1):
# if replicon contains illegal characters
# or replicon is too short < 50 bp
# then replicon is None
if replicon is not None:
_log.info("############ Processing replicon {} ({}/{}) ############\n".format(replicon.id,
rep_no,
sequences_db_len))
integron_res, summary = find_integron_in_one_replicon(replicon, config)
if integron_res:
all_integrons.append(integron_res)
if summary:
all_summaries.append(summary)
else:
_log.warning("############ Skipping replicon {}/{} ############".format(rep_no,
sequences_db_len))
if not config.split_results:
_log.info("Merging integrons results.\n")
agg_integrons = results.merge_results(*all_integrons)
agg_summary = results.merge_results(*all_summaries)
outfile_base_name = os.path.join(config.result_dir, utils.get_name_from_path(config.input_seq_path))
merged_integron_file = outfile_base_name + ".integrons"
if not agg_integrons.empty:
agg_integrons.to_csv(merged_integron_file, sep="\t", index=False, na_rep="NA")
else:
with open(merged_integron_file, "w") as out_f:
out_f.write("# No Integron found\n")
merged_summary_file = outfile_base_name + ".summary"
if not agg_integrons.empty:
agg_summary.to_csv(merged_summary_file, sep="\t", index=False, na_rep="NA",
columns=['ID_replicon', 'ID_integron', 'complete', 'In0', 'CALIN'])
for _file in all_integrons + all_summaries:
if _file != merged_integron_file and _file != merged_summary_file:
# in special case where the merged file has the same name that a replicon result file
os.unlink(_file)
def tearDown(self):
if os.path.exists(self.out_dir) and os.path.isdir(self.out_dir):
shutil.rmtree(self.out_dir)
logger_set_level('INFO')
def tearDown(self):
if os.path.exists(self.out_dir) and os.path.isdir(self.out_dir):
shutil.rmtree(self.out_dir)
logger_set_level('INFO')
def main(args=None, loglevel=None):
"""
main entry point to integron_finder
:param str args: the arguments passed on the command line
:param loglevel: the output verbosity
:type loglevel: a positive int or a string among 'DEBUG', 'INFO', 'WARNING', 'ERROR', 'CRITICAL'
"""
global _log
args = sys.argv[1:] if args is None else args
config = parse_args(args)
###################################
# Prepare directories for results #
###################################
# need to create directory before to init logger
# as we write log in integron_finder.out in this dir
if not os.path.exists(config.outdir):
os.mkdir(config.outdir)
else:
if not os.path.isdir(config.outdir):
msg = "outdir '{}' already exists and is not a directory".format(
config.outdir)
# _log.critical(msg)
# we can not log it because logger are not initialized yet.
raise IsADirectoryError(msg)
if not os.path.exists(config.result_dir):
os.mkdir(config.result_dir)
else:
if not os.path.isdir(config.result_dir):
msg = "result dir '{}' already exists and is not a directory".format(
config.outdir)
# _log.critical(msg)
# we can not log it because logger are not initialized yet.
raise IsADirectoryError(msg)
elif not os.access(config.result_dir, os.W_OK):
msg = "result dir '{}' already exists and is not writable".format(
config.outdir)
# _log.critical(msg)
# we can not log it because logger are not initialized yet.
raise PermissionError(msg)
####################
# init the loggers #
####################
log_file = os.path.join(config.result_dir, 'integron_finder.out')
integron_finder.init_logger(log_file=log_file, out=not config.mute)
_log = colorlog.getLogger('integron_finder')
if not loglevel:
# logs are specify from args options
logger_set_level(config.log_level)
else:
# used by unit tests to mute or unmute logs
logger_set_level(loglevel)
#######################################
# do last config check before running #
#######################################
if config.cmsearch is None:
msg = """cannot find 'cmsearch' in PATH.
Please install infernal package or setup 'cmsearch' binary path with --cmsearch option"""
_log.critical(msg)
raise RuntimeError(msg)
if config.hmmsearch is None:
msg = """cannot find 'hmmsearch' in PATH.
Please install hmmer package or setup 'hmmsearch' binary path with --hmmsearch option"""
_log.critical(msg)
raise RuntimeError(msg)
if config.prodigal is None:
msg = """cannot find 'prodigal' in PATH.
Please install prodigal package or setup 'prodigal' binary path with --prodigal option"""
_log.critical(msg)
raise RuntimeError(msg)
################
# print Header #
################
log_header = colorlog.getLogger('integron_finder.header')
logging = colorlog.logging.logging
handlers = []
header_log_file = logging.FileHandler(log_file)
handlers.append(header_log_file)
if not config.mute:
header_stream = colorlog.StreamHandler(sys.stdout)
handlers.append(header_stream)
formatter = colorlog.ColoredFormatter("%(message)s")
for h in handlers:
h.setFormatter(formatter)
log_header.addHandler(h)
log_header.setLevel(colorlog.logging.logging.INFO)
log_header.propagate = False
log_header.info(header(args))
with utils.FastaIterator(
config.input_seq_path,
dist_threshold=config.distance_threshold) as sequences_db:
################
# set topology #
################
default_topology = 'circ' if len(sequences_db) == 1 else 'lin'
if config.linear:
default_topology = 'lin'
elif config.circular:
default_topology = 'circ'
# the both options are mutually exclusive
topologies = Topology(default_topology,
topology_file=config.topology_file)
# allow sequences_db to inject topology information
# in seq.topology attribute
sequences_db.topologies = topologies
##############
# do the job #
##############
sequences_db_len = len(sequences_db)
all_integrons = []
all_summaries = []
for rep_no, replicon in enumerate(sequences_db, 1):
# if replicon contains illegal characters
# or replicon is too short < 50 bp
# then replicon is None
if replicon is not None:
_log.info(
"############ Processing replicon {} ({}/{}) ############\n"
.format(replicon.id, rep_no, sequences_db_len))
integron_res, summary = find_integron_in_one_replicon(
replicon, config)
if integron_res:
all_integrons.append(integron_res)
if summary:
all_summaries.append(summary)
else:
_log.warning(
"############ Skipping replicon {}/{} ############".format(
rep_no, sequences_db_len))
if not config.split_results:
_log.info("Merging integrons results.\n")
agg_integrons = results.merge_results(*all_integrons)
agg_summary = results.merge_results(*all_summaries)
outfile_base_name = os.path.join(
config.result_dir, utils.get_name_from_path(config.input_seq_path))
merged_integron_file = outfile_base_name + ".integrons"
if not agg_integrons.empty:
agg_integrons.to_csv(merged_integron_file,
sep="\t",
index=False,
na_rep="NA")
else:
with open(merged_integron_file, "w") as out_f:
out_f.write("# No Integron found\n")
merged_summary_file = outfile_base_name + ".summary"
if not agg_integrons.empty:
agg_summary.to_csv(merged_summary_file,
sep="\t",
index=False,
na_rep="NA",
columns=[
'ID_replicon', 'ID_integron', 'complete',
'In0', 'CALIN'
])
for _file in all_integrons + all_summaries:
if _file != merged_integron_file and _file != merged_summary_file:
# in special case where the merged file has the same name that a replicon result file
os.unlink(_file)
def setUp(self):
logger_set_level('INFO')
self.tmp_dir = os.path.join(tempfile.gettempdir(), 'tmp_test_integron_finder')
if os.path.isdir(self.tmp_dir):
shutil.rmtree(self.tmp_dir)
os.makedirs(self.tmp_dir)
