def test_find_integrase_gembase(self):
cfg = Config(self.args)
self.args.gembase = True
cfg._prefix_data = os.path.join(os.path.dirname(__file__), 'data')
replicon_name = 'acba.007.p01.13'
replicon_path = self.find_data(
os.path.join('Replicons', replicon_name + '.fst'))
topologies = Topology('lin')
with FastaIterator(replicon_path) as sequences_db:
sequences_db.topologies = topologies
replicon = next(sequences_db)
prot_file = os.path.join(self.tmp_dir, replicon_name + ".prt")
shutil.copyfile(
self.find_data(os.path.join('Proteins', replicon.id + ".prt")),
prot_file)
integrase.find_integrase(replicon.id, prot_file, self.tmp_dir, cfg)
for suffix in ('_intI.res', '_intI_table.res', '_phage_int.res',
'_phage_int_table.res'):
res = os.path.join(self.tmp_dir, replicon.id + suffix)
self.assertTrue(os.path.exists(res))
def test_find_integrase_no_gembase_no_protfile_no_prodigal(self):
try:
self.args.hmmsearch = 'foo'
self.args.gembase = False
cfg = Config(self.args)
cfg._prefix_data = os.path.join(os.path.dirname(__file__), 'data')
replicon_name = 'acba.007.p01.13'
replicon_path = self.find_data(
os.path.join('Replicons', replicon_name + '.fst'))
topologies = Topology('lin')
with FastaIterator(replicon_path) as sequences_db:
sequences_db.topologies = topologies
replicon = next(sequences_db)
len_ori = replicon.__class__.__len__
replicon.__class__.__len__ = lambda x: 500000
prot_file = os.path.join(self.tmp_dir, replicon.id + ".prt")
shutil.copyfile(
self.find_data(os.path.join('Proteins', replicon.id + ".prt")),
prot_file)
with self.assertRaises(RuntimeError) as ctx:
integrase.find_integrase(replicon.id, prot_file, self.tmp_dir,
cfg)
self.assertTrue(
re.search(
"failed : \[Errno 2\] No such file or directory: 'foo'",
str(ctx.exception)))
finally:
replicon.__class__.__len__ = len_ori
def test_find_integrase_no_gembase_no_protfile(self):
try:
cfg = Config(self.args)
self.args.gembase = False
cfg._prefix_data = os.path.join(os.path.dirname(__file__), 'data')
replicon_name = 'acba.007.p01.13'
replicon_path = self.find_data(
os.path.join('Replicons', replicon_name + '.fst'))
topologies = Topology('lin')
with FastaIterator(replicon_path) as sequences_db:
sequences_db.topologies = topologies
replicon = next(sequences_db)
len_ori = replicon.__class__.__len__
replicon.__class__.__len__ = lambda x: 500000
prot_file = os.path.join(self.tmp_dir, "foo.prt")
open(prot_file, 'w').close()
with self.catch_log():
with self.assertRaises(EmptyFileError) as ctx:
integrase.find_integrase(replicon.id, prot_file,
self.tmp_dir, cfg)
finally:
replicon.__class__.__len__ = len_ori
def test_find_integrase_no_gembase_no_protfile_short_seq(self):
try:
cfg = Config(self.args)
self.args.gembase = False
cfg._prefix_data = os.path.join(os.path.dirname(__file__), 'data')
replicon_name = 'acba.007.p01.13'
replicon_path = self.find_data(
os.path.join('Replicons', replicon_name + '.fst'))
prot_name = 'ACBA.007.P01_13'
prot_path = self.find_data(
os.path.join('Proteins', prot_name + '.prt'))
topologies = Topology('lin')
with FastaIterator(replicon_path) as sequences_db:
sequences_db.topologies = topologies
replicon = next(sequences_db)
len_ori = replicon.__class__.__len__
replicon.__class__.__len__ = lambda x: 200
prot_file = os.path.join(self.tmp_dir, replicon.id + ".prt")
shutil.copyfile(prot_path, prot_file)
integrase.find_integrase(replicon.id, prot_file, self.tmp_dir, cfg)
for suffix in ('_intI.res', '_intI_table.res', '_phage_int.res',
'_phage_int_table.res'):
res = os.path.join(self.tmp_dir, replicon.id + suffix)
self.assertTrue(os.path.exists(res))
finally:
replicon.__class__.__len__ = len_ori
def test_find_integrase_no_gembase_with_protfile_empty(self):
try:
cfg = Config(self.args)
self.args.gembase = False
cfg._prefix_data = os.path.join(os.path.dirname(__file__), 'data')
replicon_name = 'acba.007.p01.13'
replicon_path = self.find_data(
os.path.join('Replicons', replicon_name + '.fst'))
topologies = Topology('lin')
with FastaIterator(replicon_path) as sequences_db:
sequences_db.topologies = topologies
replicon = next(sequences_db)
len_ori = replicon.__class__.__len__
replicon.__class__.__len__ = lambda x: 200
prot_file = os.path.join(self.tmp_dir, replicon.id + ".prt")
open(prot_file, 'w').close()
with self.assertRaises(EmptyFileError) as ctx:
with self.catch_log():
integrase.find_integrase(replicon.id, prot_file,
self.tmp_dir, cfg)
self.assertTrue(
re.match(
"^The protein file: '.*' is empty cannot perform hmmsearch on it.$",
str(ctx.exception)))
finally:
replicon.__class__.__len__ = len_ori
def setUp(self):
"""
Define variables common to all tests
"""
replicon_name = "acba.007.p01.13"
self.replicon_path = self.find_data(os.path.join('Replicons', replicon_name + '.fst'))
topologies = Topology('lin')
with FastaIterator(self.replicon_path) as sequences_db:
sequences_db.topologies = topologies
self.replicon = next(sequences_db)
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)
# Resfams is too big to bee in tests/data
# search directly in data
self.hmm_files = [os.path.normpath(
os.path.join(os.path.dirname(__file__), "..", "data", "Functional_annotation", "Resfams.hmm")
)]
# Define integron_finder variables
args = argparse.Namespace()
args.gembase = False
args.annot_parser_name = None
args.hmmsearch = distutils.spawn.find_executable("hmmsearch")
args.prodigal = distutils.spawn.find_executable("prodigal")
args.cpu = 1
args.out_dir = self.tmp_dir
self.cfg = Config(args)
self.cfg._prefix_data = os.path.join(os.path.dirname(__file__), 'data')
prot_dir = os.path.join(self.tmp_dir, 'Proteins')
os.makedirs(prot_dir)
self.prot_file = os.path.join(prot_dir, self.replicon.name + ".prt")
shutil.copyfile(self.find_data(os.path.join('Proteins', self.replicon.id + ".prt")), self.prot_file)
self.prot_db = ProdigalDB(self.replicon, self.cfg, prot_file=self.prot_file)
self.exp_files = ["{}{}".format(self.replicon.id, suffix) for suffix in ("_Resfams_fa_table.res",
"_intI_table.res",
"_phage_int_table.res",
"_Resfams_fa.res",
"_intI.res",
"_phage_int.res",
"_subseqprot.tmp")]
self.exp_files = [os.path.join(self.tmp_dir, file) for file in self.exp_files]
self.prot_dtype = {"pos_beg": 'int',
"pos_end": 'int',
"strand": 'int',
"evalue": 'float',
"type_elt": 'str',
"annotation": 'str',
"model": 'str',
"distance_2attC": 'float'}
# Run prodigal to find CDS on replicon (and run hmmsearch on integrase (2 profiles))
self.integrases = find_integrase(self.replicon.id, self.prot_file, self.tmp_dir, self.cfg)
annotation.call = self.mute_call(_annot_call_ori)
def test_find_integrase_gembase_hmmer_error(self):
self.args.gembase = True
self.args.cpu = 'foo'
cfg = Config(self.args)
cfg._prefix_data = os.path.join(os.path.dirname(__file__), 'data')
replicon_name = 'acba.007.p01.13'
replicon_path = os.path.join(self._data_dir, 'Replicons',
replicon_name + '.fst')
topologies = Topology('lin')
with FastaIterator(replicon_path) as sequences_db:
sequences_db.topologies = topologies
replicon = next(sequences_db)
prot_file = os.path.join(self.tmp_dir, replicon.id + ".prt")
shutil.copyfile(
os.path.join(self._data_dir, 'Proteins', replicon.id + ".prt"),
prot_file)
with self.assertRaises(RuntimeError) as ctx:
integrase.find_integrase(replicon.id, prot_file, self.tmp_dir, cfg)
self.assertTrue(str(ctx.exception).endswith('failed return code = 1'))
def test_find_integrase_gembase_no_hmmer_no_replicon(self):
self.args.gembase = True
self.args.hmmsearch = 'foo'
cfg = Config(self.args)
cfg._prefix_data = os.path.join(os.path.dirname(__file__), 'data')
replicon_name = 'acba.007.p01.13'
replicon_path = os.path.join(self._data_dir, 'Replicons',
replicon_name + '.fst')
topologies = Topology('lin')
with FastaIterator(replicon_path) as sequences_db:
sequences_db.topologies = topologies
replicon = next(sequences_db)
prot_file = os.path.join(self.tmp_dir, replicon.id + ".prt")
with self.catch_log():
with self.assertRaises(RuntimeError) as ctx:
integrase.find_integrase(replicon.id, prot_file, self.tmp_dir,
cfg)
self.assertEqual(
"The protein file: '{}' does not exists cannot perform hmmsearch on it."
.format(prot_file), str(ctx.exception))
def find_integron_in_one_replicon(replicon, config):
"""
scan replicon for integron.
* presence of integrase
* presence of attC sites
* presence of promoters and attI sites
depending on the configuration
* perform functional annotation
produce a file containing presence of putative integrons
depending on configuration
* produce genbank file with replicon and annotations with integrons
* produce schema of replicon with integrons (in pdf)
:param replicon: the replicon to analyse.
:type replicon: a :class:`Bio.SeqRecord` object.
:param config: The configuration
:type config: a :class:`integron_finder.config.Config` object.
:returns: the path to the integron file (<replicon_id>.integrons)
and the summary file (<replicon_id.summary>).
if there is no integron the summary file is None
:rtype: tuple (str integron_file, str summary_file) or (str integron_file, None)
"""
result_tmp_dir = config.tmp_dir(replicon.id)
try:
os.mkdir(result_tmp_dir)
except OSError:
pass
tmp_replicon_path = os.path.join(result_tmp_dir, replicon.id + '.fst')
SeqIO.write(replicon, tmp_replicon_path, "fasta")
# create attr path
# used to generate protein file with prodigal
replicon.path = tmp_replicon_path
# func_annot_path is the canonical path for Functional_annotation
# path_func_annot is the path provide on the command line
if config.func_annot and not config.no_proteins and not config.path_func_annot:
if os.path.exists('bank_hmm'):
fa_hmm = scan_hmm_bank('bank_hmm')
elif os.path.exists(config.func_annot_path):
fa_hmm = scan_hmm_bank(config.func_annot_path)
else:
raise IntegronError("the dir '{}' neither 'bank_hmm' exists, specify the location of hmm "
"profile with --path-func-annot option".format(config.func_annot_path))
is_func_annot = True
elif config.path_func_annot and config.no_proteins is False:
fa_hmm = scan_hmm_bank(config.path_func_annot)
is_func_annot = True
else:
is_func_annot = False
if is_func_annot and not fa_hmm:
_log.warning("No hmm profiles for functional annotation detected, skip functional annotation step.")
if config.gembase_path:
protein_db = GembaseDB(replicon, config, gembase_path=config.gembase_path)
elif config.gembase:
protein_db = GembaseDB(replicon, config)
else:
protein_db = ProdigalDB(replicon, config)
##################
# Default search #
##################
intI_file = os.path.join(result_tmp_dir, replicon.id + "_intI.res")
phageI_file = os.path.join(result_tmp_dir, replicon.id + "_phage_int.res")
attC_default_file = os.path.join(result_tmp_dir, replicon.id + "_attc_table.res")
try:
if not config.no_proteins:
if not os.path.isfile(intI_file) or not os.path.isfile(phageI_file):
find_integrase(replicon.id, protein_db.protfile, result_tmp_dir, config)
_log.info("Starting Default search ... :")
if not os.path.isfile(attC_default_file):
# find attc with cmsearch
find_attc(tmp_replicon_path, replicon.name, config.cmsearch, result_tmp_dir, config.model_attc_path,
incE=config.evalue_attc,
cpu=config.cpu)
_log.info("Default search done... : ")
integrons = find_integron(replicon, protein_db, attC_default_file, intI_file, phageI_file, config)
#########################
# Search with local_max #
#########################
if config.local_max:
_log.info("Starting search with local_max...:")
if not os.path.isfile(os.path.join(result_tmp_dir, "integron_max.pickle")):
circular = True if replicon.topology == 'circ' else False
integron_max = find_attc_max(integrons, replicon, config.distance_threshold,
config.model_attc_path,
max_attc_size=config.max_attc_size,
min_attc_size=config.min_attc_size,
circular=circular, out_dir=result_tmp_dir,
cpu=config.cpu,
evalue_attc=config.evalue_attc)
integron_max.to_pickle(os.path.join(result_tmp_dir, "integron_max.pickle"))
_log.info("Search with local_max done... :")
else:
integron_max = pd.read_pickle(os.path.join(result_tmp_dir, "integron_max.pickle"))
integron_max = integron_max[(integron_max.evalue < config.evalue_attc) &
(abs(integron_max.pos_end - integron_max.pos_beg) < config.max_attc_size) &
(config.min_attc_size < abs(integron_max.pos_end - integron_max.pos_beg))]
_log.info("Search with local_max was already done, continue... :")
integrons = find_integron(replicon, protein_db, integron_max, intI_file, phageI_file, config)
##########################
# Add promoters and attI #
##########################
for integron in integrons:
integron_type = integron.type()
if integron_type != "In0": # complete & CALIN
if not config.no_proteins:
_log.info("Adding proteins ... :")
integron.add_proteins(protein_db)
if config.promoter_attI:
_log.info("Adding promoters and attI ... :")
if integron_type == "complete":
integron.add_promoter()
integron.add_attI()
elif integron_type == "In0":
integron.add_attI()
integron.add_promoter()
#########################
# Functional annotation #
#########################
if is_func_annot and fa_hmm:
_log.info("Starting functional annotation ...:")
func_annot(integrons, replicon, protein_db, fa_hmm, config, result_tmp_dir)
#######################
# Writing out results #
#######################
_log.info("Writing out results for replicon {}".format(replicon.id))
if config.pdf:
for j, integron in enumerate(integrons, 1):
if integron.type() == "complete":
integron.draw_integron(file=os.path.join(config.result_dir, "{}_{}.pdf".format(replicon.id, j)))
base_outfile = os.path.join(config.result_dir, replicon.id)
integron_file = base_outfile + ".integrons"
_log.debug("Writing integron_file {}".format(integron_file))
if integrons:
integrons_report = results.integrons_report(integrons)
integrons_report.to_csv(integron_file, sep="\t", index=False, na_rep="NA")
summary = results.summary(integrons_report)
summary_file = base_outfile + ".summary"
summary.to_csv(summary_file, sep="\t", na_rep="NA", index=False,
columns=['ID_replicon', 'ID_integron', 'complete', 'In0', 'CALIN'])
if config.gbk:
add_feature(replicon, integrons_report, protein_db, config.distance_threshold)
SeqIO.write(replicon, os.path.join(config.result_dir, replicon.id + ".gbk"), "genbank")
else:
with open(integron_file, "w") as out_f:
out_f.write("# No Integron found\n")
summary_file = None
except integron_finder.EmptyFileError as err:
_log.warning('############ Skip replicon {} ############'.format(replicon.name))
integron_file = ''
summary_file = ''
#########################
# clean temporary files #
#########################
if not config.keep_tmp:
try:
shutil.rmtree(result_tmp_dir)
except Exception as err:
_log.warning("Cannot remove temporary results : '{} : {}'".format(result_tmp_dir, str(err)))
return integron_file, summary_file
def find_integron_in_one_replicon(replicon, config):
"""
scan replicon for integron.
* presence of integrase
* presence of attC sites
* presence of promoters and attI sites
depending on the configuration
* perform functional annotation
produce a file containing presence of putative integrons
depending on configuration
* produce genbank file with replicon and annotations with integrons
* produce schema of replicon with integrons (in pdf)
:param replicon: the replicon to analyse.
:type replicon: a :class:`Bio.SeqRecord` object.
:param config: The configuration
:type config: a :class:`integron_finder.config.Config` object.
:returns: the path to the integron file (<replicon_id>.integrons)
and the summary file (<replicon_id.summary>).
if there is no integron the summary file is None
:rtype: tuple (str integron_file, str summary_file) or (str integron_file, None)
"""
result_tmp_dir = config.tmp_dir(replicon.id)
try:
os.mkdir(result_tmp_dir)
except OSError:
pass
tmp_replicon_path = os.path.join(result_tmp_dir, replicon.id + '.fst')
SeqIO.write(replicon, tmp_replicon_path, "fasta")
# create attr path
# used to generate protein file with prodigal
replicon.path = tmp_replicon_path
# func_annot_path is the canonical path for Functional_annotation
# path_func_annot is the path provide on the command line
if config.func_annot and not config.no_proteins and not config.path_func_annot:
if os.path.exists('bank_hmm'):
fa_hmm = scan_hmm_bank('bank_hmm')
elif os.path.exists(config.func_annot_path):
fa_hmm = scan_hmm_bank(config.func_annot_path)
else:
raise IntegronError(
"the dir '{}' neither 'bank_hmm' exists, specify the location of hmm "
"profile with --path-func-annot option".format(
config.func_annot_path))
is_func_annot = True
elif config.path_func_annot and config.no_proteins is False:
fa_hmm = scan_hmm_bank(config.path_func_annot)
is_func_annot = True
else:
is_func_annot = False
if is_func_annot and not fa_hmm:
_log.warning(
"No hmm profiles for functional annotation detected, skip functional annotation step."
)
if config.gembase_path:
protein_db = GembaseDB(replicon,
config,
gembase_path=config.gembase_path)
elif config.gembase:
protein_db = GembaseDB(replicon, config)
else:
protein_db = ProdigalDB(replicon, config)
##################
# Default search #
##################
intI_file = os.path.join(result_tmp_dir, replicon.id + "_intI.res")
phageI_file = os.path.join(result_tmp_dir, replicon.id + "_phage_int.res")
attC_default_file = os.path.join(result_tmp_dir,
replicon.id + "_attc_table.res")
try:
if not config.no_proteins:
if not os.path.isfile(intI_file) or not os.path.isfile(
phageI_file):
find_integrase(replicon.id, protein_db.protfile,
result_tmp_dir, config)
_log.info("Starting Default search ... :")
if not os.path.isfile(attC_default_file):
# find attc with cmsearch
find_attc(tmp_replicon_path,
replicon.name,
config.cmsearch,
result_tmp_dir,
config.model_attc_path,
incE=config.evalue_attc,
cpu=config.cpu)
_log.info("Default search done... : ")
integrons = find_integron(replicon, protein_db, attC_default_file,
intI_file, phageI_file, config)
#########################
# Search with local_max #
#########################
if config.local_max:
_log.info("Starting search with local_max...:")
if not os.path.isfile(
os.path.join(result_tmp_dir, "integron_max.pickle")):
circular = True if replicon.topology == 'circ' else False
integron_max = find_attc_max(
integrons,
replicon,
config.distance_threshold,
config.model_attc_path,
max_attc_size=config.max_attc_size,
min_attc_size=config.min_attc_size,
circular=circular,
out_dir=result_tmp_dir,
cpu=config.cpu,
evalue_attc=config.evalue_attc)
integron_max.to_pickle(
os.path.join(result_tmp_dir, "integron_max.pickle"))
_log.info("Search with local_max done... :")
else:
integron_max = pd.read_pickle(
os.path.join(result_tmp_dir, "integron_max.pickle"))
integron_max = integron_max[
(integron_max.evalue < config.evalue_attc)
& (abs(integron_max.pos_end -
integron_max.pos_beg) < config.max_attc_size) &
(config.min_attc_size <
abs(integron_max.pos_end - integron_max.pos_beg))]
_log.info(
"Search with local_max was already done, continue... :")
integrons = find_integron(replicon, protein_db, integron_max,
intI_file, phageI_file, config)
##########################
# Add promoters and attI #
##########################
for integron in integrons:
integron_type = integron.type()
if integron_type != "In0": # complete & CALIN
if not config.no_proteins:
_log.info("Adding proteins ... :")
integron.add_proteins(protein_db)
if config.promoter_attI:
_log.info("Adding promoters and attI ... :")
if integron_type == "complete":
integron.add_promoter()
integron.add_attI()
elif integron_type == "In0":
integron.add_attI()
integron.add_promoter()
#########################
# Functional annotation #
#########################
if is_func_annot and fa_hmm:
_log.info("Starting functional annotation ...:")
func_annot(integrons, replicon, protein_db, fa_hmm, config,
result_tmp_dir)
#######################
# Writing out results #
#######################
_log.info("Writing out results for replicon {}".format(replicon.id))
if config.pdf:
for j, integron in enumerate(integrons, 1):
if integron.type() == "complete":
integron.draw_integron(file=os.path.join(
config.result_dir, "{}_{}.pdf".format(replicon.id, j)))
base_outfile = os.path.join(config.result_dir, replicon.id)
integron_file = base_outfile + ".integrons"
_log.debug("Writing integron_file {}".format(integron_file))
if integrons:
integrons_report = results.integrons_report(integrons)
integrons_report.to_csv(integron_file,
sep="\t",
index=False,
na_rep="NA")
summary = results.summary(integrons_report)
summary_file = base_outfile + ".summary"
summary.to_csv(summary_file,
sep="\t",
na_rep="NA",
index=False,
columns=[
'ID_replicon', 'ID_integron', 'complete', 'In0',
'CALIN'
])
if config.gbk:
add_feature(replicon, integrons_report, protein_db,
config.distance_threshold)
SeqIO.write(
replicon,
os.path.join(config.result_dir, replicon.id + ".gbk"),
"genbank")
else:
with open(integron_file, "w") as out_f:
out_f.write("# No Integron found\n")
summary_file = None
except integron_finder.EmptyFileError as err:
_log.warning('############ Skip replicon {} ############'.format(
replicon.name))
integron_file = ''
summary_file = ''
#########################
# clean temporary files #
#########################
if not config.keep_tmp:
try:
shutil.rmtree(result_tmp_dir)
except Exception as err:
_log.warning("Cannot remove temporary results : '{} : {}'".format(
result_tmp_dir, str(err)))
return integron_file, summary_file