def test_search_attc_uniq(self):
"""
Test that it finds a unique attc array when giving a table with 3 attC sites
on the same strand and separated by less than 4kb each.
"""
attc_file = os.path.join("tests", "data", "Results_Integron_Finder_" + self.rep_name,
"other", self.rep_name + "_attc_table.res")
# Construct attC dataframe (read from infernal file)
attc_df = integron_finder.read_infernal(attc_file)
# search attC arrays, keeping palindromes
# 2 attc sites are in the same array if they are on the same strand, and separated by
# a distance less than 4kb
attc_array = integron_finder.search_attc(attc_df, True)
self.assertEqual(len(attc_array), 1)
# Construct expected output:
attc_res = pd.DataFrame(columns=["Accession_number", "cm_attC", "cm_debut", "cm_fin", "pos_beg", "pos_end", "sens", "evalue"], dtype='int')
attc_res = attc_res.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 17825,
"pos_end": 17884, "sens": "-", "evalue": 1e-9},
ignore_index=True)
attc_res = attc_res.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 19080,
"pos_end": 19149, "sens": "-", "evalue": 1e-4},
ignore_index=True)
attc_res = attc_res.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 19618,
"pos_end": 19726, "sens": "-", "evalue": 1.1e-7},
ignore_index=True)
# convert positions to int
intcols = ["cm_debut", "cm_fin", "pos_beg", "pos_end"]
attc_res[intcols] = attc_res[intcols].astype(int)
pdt.assert_frame_equal(attc_res, attc_array[0])
def test_filter_evalue_thres(self):
"""
Test that the filter by a maximum attc size works.
"""
filename = os.path.join("tests", "data",
"Results_Integron_Finder_" + self.rep_name,
"other", self.rep_name + "_attc_table.res")
df = integron_finder.read_infernal(filename, evalue=1e-8)
expect = pd.DataFrame(columns=[
"Accession_number", "cm_attC", "cm_debut", "cm_fin", "pos_beg",
"pos_end", "sens", "evalue"
])
expect = expect.append(
{
"Accession_number": self.rep_name,
"cm_attC": "attC_4",
"cm_debut": 1,
"cm_fin": 47,
"pos_beg": 17825,
"pos_end": 17884,
"sens": "-",
"evalue": 1e-9
},
ignore_index=True)
# convert positions to int
intcols = ["cm_debut", "cm_fin", "pos_beg", "pos_end"]
expect[intcols] = expect[intcols].astype(int)
pdt.assert_frame_equal(df, expect)
def test_no_total_cm_match_strandm(self):
"""
Test that when the model did not completely match on the sequence,
the start and end positions of hit are well recalculated. All hits are on strand -
"""
filename = os.path.join("tests", "data",
"Results_Integron_Finder_" + self.rep_name,
"other",
self.rep_name + "_attc_table-partialm.res")
df = integron_finder.read_infernal(filename)
expect = pd.DataFrame(columns=[
"Accession_number", "cm_attC", "cm_debut", "cm_fin", "pos_beg",
"pos_end", "sens", "evalue"
])
expect = expect.append(
{
"Accession_number": self.rep_name,
"cm_attC": "attC_4",
"cm_debut": 1,
"cm_fin": 40,
"pos_beg": 17818,
"pos_end": 17884,
"sens": "-",
"evalue": 1e-9
},
ignore_index=True)
expect = expect.append(
{
"Accession_number": self.rep_name,
"cm_attC": "attC_4",
"cm_debut": 1,
"cm_fin": 47,
"pos_beg": 19080,
"pos_end": 19149,
"sens": "-",
"evalue": 1e-4
},
ignore_index=True)
expect = expect.append(
{
"Accession_number": self.rep_name,
"cm_attC": "attC_4",
"cm_debut": 10,
"cm_fin": 47,
"pos_beg": 19618,
"pos_end": 19735,
"sens": "-",
"evalue": 1.1e-7
},
ignore_index=True)
# convert positions to int
intcols = ["cm_debut", "cm_fin", "pos_beg", "pos_end"]
expect[intcols] = expect[intcols].astype(int)
pdt.assert_frame_equal(df, expect)
def test_generate_df(self):
"""
Test that if the infernal file exists and there are hits, it returns the
dataframe corresponding to it.
"""
filename = os.path.join("tests", "data",
"Results_Integron_Finder_" + self.rep_name,
"other", self.rep_name + "_attc_table.res")
df = integron_finder.read_infernal(filename)
expect = pd.DataFrame(columns=[
"Accession_number", "cm_attC", "cm_debut", "cm_fin", "pos_beg",
"pos_end", "sens", "evalue"
])
expect = expect.append(
{
"Accession_number": self.rep_name,
"cm_attC": "attC_4",
"cm_debut": 1,
"cm_fin": 47,
"pos_beg": 17825,
"pos_end": 17884,
"sens": "-",
"evalue": 1e-9
},
ignore_index=True)
expect = expect.append(
{
"Accession_number": self.rep_name,
"cm_attC": "attC_4",
"cm_debut": 1,
"cm_fin": 47,
"pos_beg": 19080,
"pos_end": 19149,
"sens": "-",
"evalue": 1e-4
},
ignore_index=True)
expect = expect.append(
{
"Accession_number": self.rep_name,
"cm_attC": "attC_4",
"cm_debut": 1,
"cm_fin": 47,
"pos_beg": 19618,
"pos_end": 19726,
"sens": "-",
"evalue": 1.1e-7
},
ignore_index=True)
# convert positions to int
intcols = ["cm_debut", "cm_fin", "pos_beg", "pos_end"]
expect[intcols] = expect[intcols].astype(int)
pdt.assert_frame_equal(df, expect)
def test_search_attc_empty(self):
"""
Test that when there are no attC sites detected, the attc array is empty.
"""
attc_file = os.path.join("tests", "data", "Results_Integron_Finder_" + self.rep_name,
"other", self.rep_name + "_attc_table-empty.res")
# Construct attC dataframe (read from infernal file)
attc_df = integron_finder.read_infernal(attc_file)
attc_array = integron_finder.search_attc(attc_df, True)
self.assertEqual(len(attc_array), 0)
attc_res = []
self.assertEqual(attc_array, attc_res)
def test_nofile(self):
"""
Test that the function returns an empty dataframe if the given infernal file does not
exist.
"""
filename = "infernal.txt"
df = integron_finder.read_infernal(filename)
expect = pd.DataFrame(columns=[
"Accession_number", "cm_attC", "cm_debut", "cm_fin", "pos_beg",
"pos_end", "sens", "evalue"
])
pdt.assert_frame_equal(df, expect)
def test_evalue_thres(self):
"""
Test that if the infernal file exists and there are hits, but
the given evalue threshold is smaller than the hits thresholds:
no hit kept, should return an empty dataframe.
"""
filename = os.path.join("tests", "data",
"Results_Integron_Finder_" + self.rep_name,
"other", self.rep_name + "_attc_table.res")
df = integron_finder.read_infernal(filename, evalue=1e-10)
expect = pd.DataFrame(columns=[
"Accession_number", "cm_attC", "cm_debut", "cm_fin", "pos_beg",
"pos_end", "sens", "evalue"
])
pdt.assert_frame_equal(df, expect)
def test_nohit(self):
"""
Test that if the infernal file exists but there is no hit
inside, it returns an empty dataframe.
"""
filename = os.path.join("tests", "data",
"Results_Integron_Finder_" + self.rep_name,
"other",
self.rep_name + "_attc_table-empty.res")
df = integron_finder.read_infernal(filename)
expect = pd.DataFrame(columns=[
"Accession_number", "cm_attC", "cm_debut", "cm_fin", "pos_beg",
"pos_end", "sens", "evalue"
])
pdt.assert_frame_equal(df, expect)
def test_find_integron_attC_is_df(self):
replicon_name = 'acba.007.p01.13'
replicon_path = os.path.join(self._data_dir, 'Replicons',
replicon_name + '.fst')
attc_file = os.path.join(
self._data_dir,
'Results_Integron_Finder_acba.007.p01.13/other/acba.007.p01.13_attc_table.res'
)
intI_file = os.path.join(
self._data_dir,
'Results_Integron_Finder_acba.007.p01.13/other/acba.007.p01.13_intI.res'
)
phageI_file = os.path.join(
self._data_dir,
'Results_Integron_Finder_acba.007.p01.13/other/acba.007.p01.13_phage_int.res'
)
args = argparse.Namespace
args.no_proteins = True
args.keep_palindromes = True
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_finder.evalue_attc = 1.
integron_finder.max_attc_size = 200
integron_finder.min_attc_size = 40
integron_finder.length_cm = 47 # length in 'CLEN' (value for model attc_4.cm)
integron_finder.DISTANCE_THRESHOLD = 4000 # (4kb at least between 2 different arrays)
integron_finder.model_attc_name = integron_finder.MODEL_attc.split(
"/")[-1].split(".cm")[0]
attc_file = integron_finder.read_infernal(
attc_file,
evalue=integron_finder.evalue_attc,
size_max_attc=integron_finder.max_attc_size,
size_min_attc=integron_finder.min_attc_size)
with self.catch_output() as (out, err):
integrons = integron_finder.find_integron(replicon_name, attc_file,
intI_file, phageI_file)
self.assertEqual(err.getvalue().strip(), "")
self.assertEqual(
out.getvalue().strip(), """In replicon acba.007.p01.13, there are:
- 0 complete integron(s) found with a total 0 attC site(s)
- 1 CALIN element(s) found with a total of 3 attC site(s)
- 0 In0 element(s) found with a total of 0 attC site""")
self.assertEqual(len(integrons), 1)
integron = integrons[0]
self.assertEqual(integron.ID_replicon, replicon_name)
exp = pd.DataFrame(
{
'annotation': ['attC'] * 3,
'distance_2attC': [np.nan, 1196.0, 469.0],
'evalue': [1.000000e-09, 1.000000e-04, 1.100000e-07],
'model': ['attc_4'] * 3,
'pos_beg': [17825, 19080, 19618],
'pos_end': [17884, 19149, 19726],
'strand': [-1, -1, -1],
'type_elt': 'attC'
},
columns=self.columns,
index=['attc_001', 'attc_002', 'attc_003'])
pdt.assert_frame_equal(integron.attC, exp)
exp = pd.DataFrame(columns=self.columns)
exp = exp.astype(dtype=self.dtype)
pdt.assert_frame_equal(integron.integrase, exp)
pdt.assert_frame_equal(integron.promoter, exp)
pdt.assert_frame_equal(integron.attI, exp)
pdt.assert_frame_equal(integron.proteins, exp)
def test_search_attc_dist_diff_strand(self):
"""
Test that it finds a size 3 attc array when giving a table with:
- 3 attC sites on the same strand (-) and separated by less than 4 kb
- 2 other attC sites separated by less than 4kb but on the other strand (+)
- 1 other attC site , also on strand +, but separated by more than 4kb.
"""
attc_file = os.path.join("tests", "data", "Results_Integron_Finder_" + self.rep_name,
"other", self.rep_name + "_attc_table.res")
# Construct attC dataframe (read from infernal file)
attc_df = integron_finder.read_infernal(attc_file)
# Add another attC at more than 4kb, same strand
attc_df = attc_df.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 15800,
"pos_end": 16000, "sens": "+", "evalue": 1e-3},
ignore_index=True)
attc_df = attc_df.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 12000,
"pos_end": 12500, "sens": "+", "evalue": 1e-3},
ignore_index=True)
attc_df = attc_df.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 7100,
"pos_end": 8200, "sens": "+", "evalue": 1e-3},
ignore_index=True)
attc_df.sort_values(["Accession_number", "pos_beg", "evalue"], inplace=True)
# search attC arrays, keeping palindromes
# 2 attc sites are in the same array if they are on the same strand, and separated by
# a distance less than 4kb
attc_array = integron_finder.search_attc(attc_df, True)
self.assertEqual(len(attc_array), 3)
# Construct expected outputs:
attc_res = pd.DataFrame(columns=["Accession_number", "cm_attC", "cm_debut", "cm_fin",
"pos_beg", "pos_end", "sens", "evalue"])
attc_res = attc_res.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 17825,
"pos_end": 17884, "sens": "-", "evalue": 1e-9},
ignore_index=True)
attc_res = attc_res.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 19080,
"pos_end": 19149, "sens": "-", "evalue": 1e-4},
ignore_index=True)
attc_res = attc_res.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 19618,
"pos_end": 19726, "sens": "-", "evalue": 1.1e-7},
ignore_index=True)
attc_res2 = pd.DataFrame(columns=["Accession_number", "cm_attC", "cm_debut", "cm_fin",
"pos_beg", "pos_end", "sens", "evalue"])
attc_res2 = attc_res2.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 12000,
"pos_end": 12500, "sens": "+", "evalue": 1e-03},
ignore_index=True)
attc_res2 = attc_res2.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 15800,
"pos_end": 16000, "sens": "+", "evalue": 1e-03},
ignore_index=True)
attc_res3 = pd.DataFrame(columns=["Accession_number", "cm_attC", "cm_debut", "cm_fin",
"pos_beg", "pos_end", "sens", "evalue"])
attc_res3 = attc_res3.append({"Accession_number": self.rep_name, "cm_attC": "attC_4",
"cm_debut": 1, "cm_fin": 47, "pos_beg": 7100,
"pos_end": 8200, "sens": "+", "evalue": 1e-03},
ignore_index=True)
# convert positions to int
intcols = ["cm_debut", "cm_fin", "pos_beg", "pos_end"]
attc_res[intcols] = attc_res[intcols].astype(int)
attc_res2[intcols] = attc_res2[intcols].astype(int)
attc_res3[intcols] = attc_res3[intcols].astype(int)
pdt.assert_frame_equal(attc_res, attc_array[2])
pdt.assert_frame_equal(attc_res2, attc_array[1])
pdt.assert_frame_equal(attc_res3, attc_array[0])