def test_invalid_same_variant_defined_in_two_rows_in_hgvs_pro(self):
hgvs = generate_hgvs(prefix="p")
data = "{},{}\n{},1.0\n{},1.0".format(self.HGVS_PRO_COL, "count", hgvs,
hgvs)
dataset = MaveDataset.for_counts(StringIO(data))
dataset.validate()
self.assertFalse(dataset.is_valid)
self.assertEqual(len(dataset.errors), 1)
print(dataset.errors)
def test_defines_same_variants(self):
tests = [
(
"{},{}\nc.1A>G,0.0".format(self.HGVS_NT_COL, self.SCORE_COL),
"{},count\nc.1A>G,0.0".format(self.HGVS_NT_COL),
True,
),
(
"{},{}\nc.1A>G,0.0".format(self.HGVS_NT_COL, self.SCORE_COL),
"{},count\nc.2A>G,0.0".format(self.HGVS_NT_COL),
False,
),
(
"{},{},{}\nc.1A>G,p.Ile1Val,0.0".format(
self.HGVS_NT_COL, self.HGVS_PRO_COL, self.SCORE_COL),
"{},{},count\nc.1A>G,p.Ile1Val,0.0".format(
self.HGVS_NT_COL, self.HGVS_PRO_COL),
True,
),
(
"{},{},{}\nc.1A>G,p.Ile1Val,0.0".format(
self.HGVS_NT_COL, self.HGVS_PRO_COL, self.SCORE_COL),
"{},{},count\nc.1A>G,p.Ile1Phe,0.0".format(
self.HGVS_NT_COL, self.HGVS_PRO_COL),
False,
),
# Check returns None if either dataset invalid
(
"wrong_columns,{}\nc.1A>G,0.0".format(self.SCORE_COL),
"{},count\nc.1A>G,0.0".format(self.HGVS_NT_COL),
None,
),
(
"{},{}\nc.1A>G,0.0".format(self.HGVS_NT_COL, self.SCORE_COL),
"wrong_column,count\nc.1A>G,0.0".format(),
None,
),
]
for (scores, counts, expected) in tests:
with self.subTest(msg=(scores, counts, expected)):
scores_dataset = MaveDataset.for_scores(StringIO(scores))
scores_dataset.validate()
counts_dataset = MaveDataset.for_counts(StringIO(counts))
counts_dataset.validate()
self.assertEqual(scores_dataset.match_other(counts_dataset),
expected)
def test_invalid_no_additional_columns_outside_hgvs_ones(self):
data = "{},{},{}\n{},{},{}".format(
self.HGVS_NT_COL,
self.HGVS_SPLICE_COL,
self.HGVS_PRO_COL,
generate_hgvs(prefix="g"),
generate_hgvs(prefix="c"),
generate_hgvs(prefix="p"),
)
dataset = MaveDataset.for_counts(StringIO(data))
dataset.validate()
self.assertFalse(dataset.is_valid)
self.assertEqual(len(dataset.errors), 1)
print(dataset.errors)