def test_import_cellosaurus(self, mock_api_call):
"""The cellosaurus ontology is not part of the OLS so we need to handle
it separately. We still include it because metaSRA uses its terms.
NOTE: the actual cellosaurus ontology is massive, so this VCR was
created using a trimmed-down cellosaurus ontology where I went in and
deleted a bunch of the publications and cell lines from the respective
lists. Besides this, no alterations were made to the original."""
# We shouldn't be hitting the API at all here, because we should have
# the ontology already imported
mock_api_call.return_value = "The wrong answer"
created_terms = OntologyTerm.import_entire_ontology("cvcl")
self.assertEqual(
OntologyTerm.objects.all().count(),
34, # This is the number I counted in the file
)
self.assertEqual(OntologyTerm.objects.all().count(), created_terms)
self.assertEqual(
"#W7079",
OntologyTerm.get_or_create_from_api(
"CVCL:E549").human_readable_name,
)
mock_api_call.assert_not_called()
def test_import_entire_ontology(self, mock_api_call):
# We shouldn't be hitting the API at all here, because we should have
# the ontology already imported
mock_api_call.return_value = "The wrong answer"
# I used the UO ontology here because it is much smaller than other important
# ontologies like EFO, which could take upwards of a minute to download and parse
OntologyTerm.import_entire_ontology("uo")
self.assertGreater(OntologyTerm.objects.all().count(), 0)
self.assertEqual(
"month",
OntologyTerm.get_or_create_from_api(
"UO:0000035").human_readable_name)
mock_api_call.assert_not_called()
def test_get_or_create_from_api_returns_cached(self, mock_api_call):
mock_api_call.return_value = "medulloblastoma"
self.assertEqual(
"medulloblastoma",
OntologyTerm.get_or_create_from_api(
"EFO:0002939").human_readable_name,
)
mock_api_call.assert_called_once_with("EFO:0002939")
mock_api_call.reset_mock()
self.assertEqual(
"medulloblastoma",
OntologyTerm.get_or_create_from_api(
"EFO:0002939").human_readable_name,
)
mock_api_call.assert_not_called()
def set_value(self, value):
"""This method sets the attribute value and assigns the correct
value_type. NOTE: we assume that all provided strings are ontology terms."""
if type(value) == str:
self.value_type = ONTOLOGY_TERM
# make sure that we know how to deal with this ontology term, and error out if we don't
OntologyTerm.get_or_create_from_api(value)
elif type(value) == bool:
self.value_type = BOOL
elif type(value) == int:
self.value_type = INT
elif type(value) == float:
self.value_type = FLOAT
else:
raise ValueError("Invalid metadata value type '{}'".format(type(value)))
self.value = str(value)
def test_import_cl(self, mock_api_call):
"""Try importing the CL ontology, which was updated at some point and
broke the original parsing code"""
# We shouldn't be hitting the API at all here, because we should have
# the ontology already imported
mock_api_call.return_value = "The wrong answer"
created_terms = OntologyTerm.import_entire_ontology("cl")
self.assertEqual(
OntologyTerm.objects.all().count(),
2493, # Since we are using a VCR, this number should not change until we refresh it
)
self.assertEqual(OntologyTerm.objects.all().count(), created_terms)
self.assertEqual(
"hematopoietic cell",
OntologyTerm.get_or_create_from_api(
"CL:0000988").human_readable_name,
)
mock_api_call.assert_not_called()
def get_value(self):
"""This method returns the value of this attribute using `value_type`
to convert to the same type"""
if self.value_type == ONTOLOGY_TERM:
return OntologyTerm.get_or_create_from_api(self.value)
elif self.value_type == BOOL:
return bool(self.value)
elif self.value_type == INT:
return int(self.value)
elif self.value_type == FLOAT:
return float(self.value)
else:
raise ValueError("Invalid value_type '{}'".format(self.value_type))
def test_get_or_create_from_api_fetches(self):
self.assertEqual(
"month",
OntologyTerm.get_or_create_from_api(
"UO:0000035").human_readable_name)
def test_get_or_create_from_cellosaurus_api(self):
self.assertEqual(
"LNCaP",
OntologyTerm.get_or_create_from_api(
"CVCL:0395").human_readable_name)