def setUp(self):
self.string1 = "Mycobacterium phage Trixie"
self.string2 = "Mycobacterium smegmatis"
self.string3 = "Gordonia terrae"
self.qualifier_dict = {
"organism": [self.string1],
"host": [self.string2],
"lab_host": [self.string3]
}
self.seqfeature = test_data_utils.create_1_part_seqfeature(
2, 10, 1, "source", qualifiers=self.qualifier_dict)
def setUp(self):
self.qualifier_dict = {
"locus_tag": ["SEA_L5_1"],
"translation": ["ABCDE"],
"transl_table": ["11"],
"product": [" gp3; terminase "],
"function": [" orf4; repressor "],
"note": [" ORF5; lysin "],
"gene": ["GP2"]
}
self.seqfeature = test_data_utils.create_1_part_seqfeature(
2, 10, 1, "CDS", qualifiers=self.qualifier_dict)
def test_parse_coordinates_10(self):
"""Verify non-compound location with fuzzy stop coordinate
is parsed correctly."""
seqfeature = test_data_utils.create_1_part_seqfeature(2,
10,
1,
"CDS",
fuzzy="stop")
start, stop, parts = flat_files.parse_coordinates(seqfeature)
with self.subTest():
self.assertEqual(start, 2)
with self.subTest():
self.assertEqual(stop, -1)
with self.subTest():
self.assertEqual(parts, 1)
def test_parse_cds_seqfeature_10(self):
"""Verify CDS feature is parsed with fuzzy coordinates."""
seqfeature = test_data_utils.create_1_part_seqfeature(
2, 10, 1, "CDS", fuzzy="start", qualifiers=self.qualifier_dict)
cds_ftr = flat_files.parse_cds_seqfeature(seqfeature)
with self.subTest():
self.assertEqual(cds_ftr.locus_tag, "SEA_L5_1")
with self.subTest():
self.assertEqual(cds_ftr.orientation, "F")
with self.subTest():
self.assertEqual(cds_ftr.start, -1)
with self.subTest():
self.assertEqual(cds_ftr.stop, 10)
with self.subTest():
self.assertEqual(cds_ftr.parts, 1)
with self.subTest():
self.assertEqual(cds_ftr.coordinate_format, "0_half_open")
with self.subTest():
self.assertEqual(cds_ftr.length, 18)
def setUp(self):
self.seqfeature1 = test_data_utils.create_1_part_seqfeature(
2, 10, 1, "CDS")
self.seqfeature2 = test_data_utils.create_1_part_seqfeature(
5000, 6000, 1, "tRNA")
self.seqfeature3 = test_data_utils.create_1_part_seqfeature(
1, 11000, 1, "source")
self.seqfeature7 = test_data_utils.create_1_part_seqfeature(
1, 9000, 1, "source")
# Wrap-around feature, directly copied from
# Biopython-parsed Alice flat file.
# start1 = 152829, stop1 = 153401, start2 = 0, stop2 = 4, strand = 1
self.seqfeature4 = test_data_utils.get_alice_cds_252_seqfeature()
self.seqfeature5 = test_data_utils.create_1_part_seqfeature(
9, 50, -1, "CDS")
self.seqfeature6 = test_data_utils.create_1_part_seqfeature(
9, 30, 1, "CDS")
self.feature_list = [
self.seqfeature1, self.seqfeature2, self.seqfeature3,
self.seqfeature4, self.seqfeature5, self.seqfeature6,
self.seqfeature7
]
self.reference1 = test_data_utils.create_reference("Jane")
self.reference2 = test_data_utils.create_reference("Doe")
self.reference3 = test_data_utils.create_reference("Smith")
self.refs_list = [self.reference1, self.reference2, self.reference3]
self.description = "Mycobacterium phage L5 complete genome"
self.organism = "Gordonia phage KatherineG"
self.source = "Streptomyces phage phiC31"
self.date = "23-JAN-2014"
self.annotation_dict = {
"accessions": [" ABC123.1 "],
"source": self.source,
"organism": self.organism,
"references": self.refs_list,
"date": self.date
}
self.record = SeqRecord(seq=Seq("atgc"),
id="OPQ123.1",
name="XYZ123",
annotations=self.annotation_dict,
description=self.description,
features=self.feature_list)
self.filepath = Path("/path/to/file/Phage_ZZZ.gb")
self.exp_date = datetime.strptime(self.date, '%d-%b-%Y')
def test_create_seqfeature_dictionary_3(self):
"""Verify feature dictionary is constructed correctly with
several different feature types."""
ftr1 = test_data_utils.create_1_part_seqfeature(2, 10, 1, "CDS")
ftr2 = test_data_utils.create_1_part_seqfeature(2, 10, 1, "tRNA")
ftr3 = test_data_utils.create_1_part_seqfeature(2, 10, 1, "tmRNA")
ftr4 = test_data_utils.create_1_part_seqfeature(2, 10, 1, "other")
ftr5 = test_data_utils.create_1_part_seqfeature(2, 10, 1, "gene")
ftr6 = test_data_utils.create_1_part_seqfeature(2, 10, 1, "CDS")
feature_list = [ftr1, ftr2, ftr3, ftr4, ftr5, ftr6]
feature_dict = flat_files.create_seqfeature_dictionary(feature_list)
with self.subTest():
self.assertEqual(len(feature_dict.keys()), 5)
with self.subTest():
self.assertEqual(len(feature_dict["CDS"]), 2)
with self.subTest():
self.assertEqual(len(feature_dict["tRNA"]), 1)
with self.subTest():
self.assertEqual(len(feature_dict["tmRNA"]), 1)
with self.subTest():
self.assertEqual(len(feature_dict["other"]), 1)
with self.subTest():
self.assertEqual(len(feature_dict["gene"]), 1)
