def test_add_proteins(self):
replicon_name = 'pssu.001.c01.13'
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._data_dir,
'{}.prt.short'.format(replicon_name))
args = argparse.Namespace()
args.gembase = False
args.annot_parser_name = None
cfg = Config(args)
integron = Integron(replicon, cfg)
data_attc = {"pos_beg": [3072863, 3073496, 3074121, 3075059, 3075593, 3076281, 3076659],
"pos_end": [3072931, 3073555, 3074232, 3075118, 3075652, 3076340, 3076718],
"strand": [-1] * 7,
"evalue": [2.5e-06, 7e-08, 6.5e-08, 3.2e-06, 4.1e-07, 1.4e-08, 4e-08],
"type_elt": ['attC'] * 7,
"annotation": ['attC'] * 7,
"model": ['attc_4'] * 7,
"distance_2attC": [np.nan, 565.0, 566.0, 827.0, 475.0, 629.0, 319.0]}
attC = pd.DataFrame(data_attc,
columns=self.columns,
index=['attc_00{}'.format(i) for i in range(len(data_attc['pos_beg']))])
attC = attC.astype(dtype=self.dtype)
integron.attC = attC
prot_db = ProdigalDB(replicon, cfg, prot_file=prot_file)
integron.add_proteins(prot_db)
exp_proteins = pd.DataFrame({'pos_beg': [3071974, 3072950, 3074243, 3076720],
'pos_end': [3072855, 3073468, 3075055, 3077511],
'strand': [-1] * 4,
'evalue': [np.nan] * 4,
'type_elt': ['protein'] * 4,
'annotation': ['protein'] * 4,
'model': ['NA'] * 4,
'distance_2attC': [np.nan] *4
},
index=['PSSU.001.C01_13_281{}'.format(i) for i in range(5, 9)],
columns=self.columns
)
exp_proteins = exp_proteins.astype(dtype=self.dtype)
pdt.assert_frame_equal(exp_proteins.sort_index(), integron.proteins.sort_index())
def test_integrons_report(self):
replicon_name = "acba.007.p01.13"
replicon_path = self.find_data(os.path.join('Replicons', replicon_name + '.fst'))
topologies = Topology('circ')
with FastaIterator(replicon_path) as sequences_db:
sequences_db.topologies = topologies
replicon = next(sequences_db)
args = argparse.Namespace()
cfg = Config(args)
cfg._args.eagle_eyes = False
cfg._args.eagle_eyes = False
cfg._args.local_max = False
integron = Integron(replicon, cfg)
columns = ['pos_beg', 'pos_end', 'strand', 'evalue', 'type_elt', 'model', 'distance_2attC', 'annotation']
dtype = {"pos_beg": 'int',
"pos_end": 'int',
"strand": 'int',
"evalue": 'float',
"type_elt": 'str',
"annotation": 'str',
"model": 'str',
"distance_2attC": 'float'}
data_integrase = {"pos_beg": 55,
"pos_end": 1014,
"strand": 1,
"evalue": 1.900000e-25,
"type_elt": "protein",
"annotation": "intI",
"model": "intersection_tyr_intI",
"distance_2attC": np.nan}
id_int = "ACBA.007.P01_13_1"
integrase = pd.DataFrame(data_integrase, columns=columns, index=[id_int])
integrase = integrase.astype(dtype=dtype)
data_attc = {"pos_beg": [17825, 19080, 19618],
"pos_end": [17884, 19149, 19726],
"strand": [-1] * 3,
"evalue": [1.000000e-09, 1.000000e-04, 1.100000e-07],
"type_elt": ["attC"] * 3,
"annotation": ["attC"] * 3,
"model": ["attc_4"] * 3,
"distance_2attC": [np.nan, 1196.0, 469.0]}
attC = pd.DataFrame(data_attc,
columns=columns,
index=['attc_00{}'.format(i) for i in range(1, 4)])
attC = attC.astype(dtype=dtype)
promoter = pd.DataFrame({'pos_beg': 25,
'pos_end': 51,
'strand': -1,
'evalue': np.nan,
'type_elt': 'Promoter',
'annotation': 'Pc_1',
'model': np.nan,
'distance_2attC': np.nan
},
index=['Pc_int1'],
columns=columns
)
promoter = promoter.astype(dtype=dtype)
proteins = pd.DataFrame({'pos_beg': [17375, 17886, 19090, 19721],
'pos_end': [17722, 18665, 19749, 20254],
'strand': [-1] * 4,
'evalue': [np.nan] * 4,
'type_elt': ['protein'] * 4,
'annotation': ['protein'] * 4,
'model': [np.nan] * 4,
'distance_2attC': [np.nan] * 4
},
index=['ACBA.007.P01_13_2{}'.format(i) for i in range(0, 4)],
columns=columns
)
proteins = proteins.astype(dtype=dtype)
integron.integrase = integrase
integron.attC = attC
integron.promoter = promoter
integron.proteins = proteins
report = results.integrons_report([integron])
exp_report = pd.read_csv(
self.find_data(os.path.join('Results_Integron_Finder_{}'.format(replicon_name),
'{}.integrons'.format(replicon_name)
)),
sep="\t"
)
exp_report = exp_report.astype(dtype=dtype)
pdt.assert_frame_equal(exp_report, report)
def test_integrons_report(self):
replicon_name = "acba.007.p01.13"
replicon_path = self.find_data(
os.path.join('Replicons', replicon_name + '.fst'))
topologies = Topology('circ')
with FastaIterator(replicon_path) as sequences_db:
sequences_db.topologies = topologies
replicon = next(sequences_db)
args = argparse.Namespace()
cfg = Config(args)
cfg._args.eagle_eyes = False
cfg._args.eagle_eyes = False
cfg._args.local_max = False
integron = Integron(replicon, cfg)
columns = [
'pos_beg', 'pos_end', 'strand', 'evalue', 'type_elt', 'model',
'distance_2attC', 'annotation'
]
dtype = {
"pos_beg": 'int',
"pos_end": 'int',
"strand": 'int',
"evalue": 'float',
"type_elt": 'str',
"annotation": 'str',
"model": 'str',
"distance_2attC": 'float'
}
data_integrase = {
"pos_beg": 55,
"pos_end": 1014,
"strand": 1,
"evalue": 1.900000e-25,
"type_elt": "protein",
"annotation": "intI",
"model": "intersection_tyr_intI",
"distance_2attC": np.nan
}
id_int = "ACBA.007.P01_13_1"
integrase = pd.DataFrame(data_integrase,
columns=columns,
index=[id_int])
integrase = integrase.astype(dtype=dtype)
data_attc = {
"pos_beg": [17825, 19080, 19618],
"pos_end": [17884, 19149, 19726],
"strand": [-1] * 3,
"evalue": [1.000000e-09, 1.000000e-04, 1.100000e-07],
"type_elt": ["attC"] * 3,
"annotation": ["attC"] * 3,
"model": ["attc_4"] * 3,
"distance_2attC": [np.nan, 1196.0, 469.0]
}
attC = pd.DataFrame(data_attc,
columns=columns,
index=['attc_00{}'.format(i) for i in range(1, 4)])
attC = attC.astype(dtype=dtype)
promoter = pd.DataFrame(
{
'pos_beg': 25,
'pos_end': 51,
'strand': -1,
'evalue': np.nan,
'type_elt': 'Promoter',
'annotation': 'Pc_1',
'model': np.nan,
'distance_2attC': np.nan
},
index=['Pc_int1'],
columns=columns)
promoter = promoter.astype(dtype=dtype)
proteins = pd.DataFrame(
{
'pos_beg': [17375, 17886, 19090, 19721],
'pos_end': [17722, 18665, 19749, 20254],
'strand': [-1] * 4,
'evalue': [np.nan] * 4,
'type_elt': ['protein'] * 4,
'annotation': ['protein'] * 4,
'model': [np.nan] * 4,
'distance_2attC': [np.nan] * 4
},
index=['ACBA.007.P01_13_2{}'.format(i) for i in range(0, 4)],
columns=columns)
proteins = proteins.astype(dtype=dtype)
integron.integrase = integrase
integron.attC = attC
integron.promoter = promoter
integron.proteins = proteins
report = results.integrons_report([integron])
exp_report = pd.read_csv(self.find_data(
os.path.join('Results_Integron_Finder_{}'.format(replicon_name),
'{}.integrons'.format(replicon_name))),
sep="\t")
exp_report = exp_report.astype(dtype=dtype)
pdt.assert_frame_equal(exp_report, report)
def test_describe(self):
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)
args = argparse.Namespace()
args.eagle_eyes = False
args.local_max = False
cfg = Config(args)
integron = Integron(replicon, cfg)
data_integrase = {
"pos_beg": 55,
"pos_end": 1014,
"strand": 1,
"evalue": 1.900000e-25,
"type_elt": "protein",
"annotation": "intI",
"model": "intersection_tyr_intI",
"distance_2attC": np.nan
}
id_int = "ACBA.007.P01_13_1"
integrase = pd.DataFrame(data_integrase,
columns=self.columns,
index=[id_int])
integrase = integrase.astype(dtype=self.dtype)
data_attc = {
"pos_beg": 10,
"pos_end": 100,
"strand": -1,
"evalue": 1.1e-07,
"type_elt": "attC",
"annotation": "attC",
"model": "attc_4",
"distance_2attC": np.nan
}
attC = pd.DataFrame(data_attc,
columns=self.columns,
index=['attc_001'])
attC = attC.astype(dtype=self.dtype)
promoter = pd.DataFrame(data_attc,
columns=self.columns,
index=['prom_001'])
promoter = promoter.astype(dtype=self.dtype)
attI = pd.DataFrame(data_attc,
columns=self.columns,
index=['attI_001'])
attI = attI.astype(dtype=self.dtype)
proteins = pd.DataFrame(data_attc,
columns=self.columns,
index=['prot_001'])
proteins = proteins.astype(dtype=self.dtype)
excp_description = pd.concat(
[integrase, attC, promoter, attI, proteins], ignore_index=False)
excp_description = excp_description.reset_index()
excp_description.columns = ["element"] + list(
excp_description.columns[1:])
excp_description["type"] = "complete"
excp_description["ID_replicon"] = replicon.id
excp_description["ID_integron"] = id(
integron) # uniq identifier of a given Integron
excp_description["default"] = "Yes"
excp_description["considered_topology"] = replicon.topology
excp_description.drop_duplicates(subset=["element"], inplace=True)
self.cfg._args.eagle_eyes = False
self.cfg._args.eagle_eyes = False
integron.integrase = integrase
integron.attC = attC
integron.promoter = promoter
integron.attI = attI
integron.proteins = proteins
recieved_description = integron.describe()
pdt.assert_frame_equal(recieved_description, excp_description)
def test_attI(self):
replicon_name = 'saen.040.p01.10'
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)
attC = pd.DataFrame(
{
'pos_beg': [104651, 105162, 106018, 107567, 108423, 108743],
'pos_end': [104710, 105221, 106087, 107626, 108482, 108832],
'strand': [-1] * 6,
'evalue': [
3.400000e-06, 7.500000e-09, 6.800000e-06, 2.800000e-07,
6.600000e-06, 1.800000e-04
],
'type_elt': ['attC'] * 6,
'annotation': ['attC'] * 6,
'model': ['attc_4'] * 6,
'distance_2attC': [np.nan, 452.0, 797.0, 1480.0, 797.0, 261.0]
},
index=['attc_00{}'.format(i) for i in range(1, 7)],
columns=self.columns)
attC = attC.astype(dtype=self.dtype)
integrase = pd.DataFrame(
{
'pos_beg': 109469,
'pos_end': 110482,
'strand': 1,
'evalue': 1.600000e-24,
'type_elt': 'protein',
'annotation': 'intI',
'model': 'intersection_tyr_intI',
'distance_2attC': np.nan
},
index=['SAEN.040.P01_10_135'],
columns=self.columns)
integrase = integrase.astype(dtype=self.dtype)
##########################################
# test promoter with attC with integrase #
##########################################
integron = Integron(replicon, self.cfg)
integron.attC = attC
integron.integrase = integrase
exp_attI = pd.DataFrame(
{
'pos_beg': [109330],
'pos_end': [109388],
'strand': [-1],
'evalue': [np.nan],
'type_elt': 'attI',
'annotation': 'attI_1',
'model': 'NA',
'distance_2attC': [np.nan]
},
index=['attI1'],
columns=self.columns)
exp_attI = exp_attI.astype(dtype=self.dtype)
integron.add_attI()
pdt.assert_frame_equal(exp_attI, integron.attI)
#############################################
# test promoter with attC without integrase #
#############################################
integron = Integron(replicon, self.cfg)
integron.attC = attC
empty_attI = pd.DataFrame(columns=self.columns)
empty_attI = empty_attI.astype(dtype=self.dtype)
integron.add_attI()
pdt.assert_frame_equal(empty_attI, integron.attI)
#############################################
# test promoter without attC with integrase #
#############################################
integron = Integron(replicon, self.cfg)
integron.integrase = integrase
integron.add_attI()
pdt.assert_frame_equal(exp_attI, integron.attI)
def test_add_promoter(self):
replicon_name = 'saen.040.p01.10'
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)
## integron_finder.SIZE_REPLICON = 148711
prot_file = os.path.join(self._data_dir, 'Proteins',
'{}.prt'.format(replicon_name))
# to test promoter we need to ad attC and integrase first
# as add_promoter use attc and integrase
attC = pd.DataFrame(
{
'pos_beg': [104651, 105162, 106018, 107567, 108423, 108743],
'pos_end': [104710, 105221, 106087, 107626, 108482, 108832],
'strand': [-1] * 6,
'evalue': [
3.400000e-06, 7.500000e-09, 6.800000e-06, 2.800000e-07,
6.600000e-06, 1.800000e-04
],
'type_elt': ['attC'] * 6,
'annotation': ['attC'] * 6,
'model': ['attc_4'] * 6,
'distance_2attC': [np.nan, 452.0, 797.0, 1480.0, 797.0, 261.0]
},
index=['attc_00{}'.format(i) for i in range(1, 7)],
columns=self.columns)
attC = attC.astype(dtype=self.dtype)
integrase = pd.DataFrame(
{
'pos_beg': 109469,
'pos_end': 110482,
'strand': 1,
'evalue': 1.600000e-24,
'type_elt': 'protein',
'annotation': 'intI',
'model': 'intersection_tyr_intI',
'distance_2attC': np.nan
},
index=['SAEN.040.P01_10_135'],
columns=self.columns)
integrase = integrase.astype(dtype=self.dtype)
##########################################
# test promoter with attC with integrase #
##########################################
integron = Integron(replicon, self.cfg)
integron.attC = attC
integron.integrase = integrase
integron.add_promoter()
exp_promoters = pd.DataFrame(
{
'pos_beg': [109413, 109439],
'pos_end': [109447, 109465],
'strand': [1, -1],
'evalue': [np.nan] * 2,
'type_elt': ['Promoter'] * 2,
'annotation': ['Pint_1', 'Pc_1'],
'model': ['NA'] * 2,
'distance_2attC': [np.nan] * 2
},
index=['P_intI1', 'Pc_int1'],
columns=self.columns)
exp_promoters = exp_promoters.astype(dtype=self.dtype)
pdt.assert_frame_equal(exp_promoters, integron.promoter)
#############################################
# test promoter with attC without integrase #
#############################################
integron = Integron(replicon, self.cfg)
integron.attC = attC
integron.add_promoter()
empty_promoter = pd.DataFrame(columns=self.columns)
empty_promoter = empty_promoter.astype(dtype=self.dtype)
pdt.assert_frame_equal(empty_promoter, integron.promoter)
#############################################
# test promoter without attC with integrase #
#############################################
integron = Integron(replicon, self.cfg)
integron.integrase = integrase
integron.add_promoter()
pdt.assert_frame_equal(exp_promoters, integron.promoter)