def test_attI(self):
replicon_name = 'saen.040.p01.10'
replicon_path = os.path.join(self._data_dir, "Replicons",
replicon_name + '.fst')
integron_finder.SIZE_REPLICON = 148711
integron_finder.PROT_file = os.path.join(
self._data_dir, 'Proteins', '{}.prt'.format(replicon_name))
integron_finder.MODEL_DIR = os.path.join(self.integron_home, "data",
"Models")
integron_finder.SEQUENCE = SeqIO.read(
replicon_path, "fasta", alphabet=Seq.IUPAC.unambiguous_dna)
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_name)
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_name)
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_name)
integron.integrase = integrase
integron.add_attI()
pdt.assert_frame_equal(exp_attI, integron.attI)
def test_describe(self):
id_replicon = "acba.007.p01.13"
integron = Integron(id_replicon)
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"] = id_replicon
excp_description["ID_integron"] = id(
integron) # uniq identifier of a given Integron
excp_description["default"] = "Yes"
excp_description.drop_duplicates(subset=["element"], inplace=True)
args = argparse.Namespace
args.eagle_eyes = False
args.local_max = False
integron_finder.args = args
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_find_attc_max_complete(self):
replicon_name = 'OBAL001.B.00005.C001'
replicon_path = os.path.join(self._data_dir, 'Replicons',
replicon_name + '.fst')
args = argparse.Namespace
args.eagle_eyes = False
args.local_max = False
integron_finder.replicon_name = replicon_name
integron_finder.SEQUENCE = SeqIO.read(
replicon_path, "fasta", alphabet=Seq.IUPAC.unambiguous_dna)
integron_finder.SIZE_REPLICON = len(integron_finder.SEQUENCE)
integron_finder.args = args
integron = Integron(replicon_name)
integrase = pd.DataFrame(
{
'pos_beg': 1545830,
'pos_end': 1546807,
'strand': -1,
'evalue': 1.100000e-21,
'type_elt': 'protein',
'annotation': 'intI',
'model': 'intersection_tyr_intI',
'distance_2attC': np.nan
},
index=['OBAL001.B.00005.C001_141'],
columns=self.columns)
integrase = integrase.astype(dtype=self.dtype)
integron.integrase = integrase
attC = pd.DataFrame(
{
'pos_beg': [1547800, 1548775],
'pos_end': [1547859, 1548834],
'strand': [1, 1],
'evalue': [0.00049, 0.00009],
'type_elt': ['attC', 'attC'],
'annotation': ['attC', 'attC'],
'model': ['attc_4', 'attc_4'],
'distance_2attC': [np.nan, 916.0]
},
index=['attc_001', 'attc_002'],
columns=self.columns)
attC = attC.astype(dtype=self.dtype)
integron.attC = attC
integrons = [integron]
max_final = integron_finder.find_attc_max(integrons)
exp = pd.DataFrame(
{
'Accession_number':
['OBAL001.B.00005.C001', 'OBAL001.B.00005.C001'],
'cm_attC': ['attC_4', 'attC_4'],
'cm_debut': [4, 1],
'cm_fin': [44, 47],
'pos_beg': [1547800, 1548775],
'pos_end': [1547859, 1548834],
'sens': ['+', '+'],
'evalue': [0.000240, 0.000045]
},
index=[0, 1],
columns=self.max_cols)
exp = exp.astype(dtype=self.max_dtype)
pdt.assert_frame_equal(max_final, exp)