def test_patients_csv_file2model_without_mhc2(self):
patients_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/balachandran_supplementary_table1_patients.tsv"
)
patients = ModelConverter.parse_patients_file(patients_file, self.hla_database)
self.assertIsNotNone(patients)
self.assertIsInstance(patients, list)
self.assertTrue(len(patients) == 58)
def _get_test_data(self):
input_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_model_file.txt")
data = pd.read_csv(input_file, sep="\t")
data = data.replace({np.nan: None})
neoantigens = ModelConverter._neoantigens_csv2objects(data)
patients_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_patient_file.txt")
patients = ModelConverter.parse_patients_file(patients_file,
self.hla_database)
patient_id = "Pt29"
return neoantigens, patients, patient_id
def _read_data(
candidate_file, json_file, patients_data, patient_id,
mhc_database: MhcDatabase) -> Tuple[List[Neoantigen], List[Patient]]:
# parse patient data
patients = ModelConverter.parse_patients_file(patients_data, mhc_database)
# parse the neoantigen candidate data
if candidate_file is not None:
neoantigens = ModelConverter.parse_candidate_file(
candidate_file, patient_id)
else:
neoantigens = ModelConverter.parse_neoantigens_json_file(json_file)
return neoantigens, patients
def setUp(self):
self.references, self.configuration = integration_test_tools.load_references(
)
self.references_mouse, self.configuration_mouse = integration_test_tools.load_references(
organism=ORGANISM_MUS_MUSCULUS)
# self.fastafile = integration_test_tools.create_temp_aminoacid_fasta_file()
# self.runner = Runner()
self.patient_id = "Pt29"
input_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_candidate_file.txt")
patients_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_patient_file.txt")
patients_file_mouse = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_patient_file_mouse.txt")
self.hla_database = self.references.get_mhc_database()
self.h2_database = self.references_mouse.get_mhc_database()
self.patients = ModelConverter.parse_patients_file(
patients_file, self.hla_database)
self.patients_mouse = ModelConverter.parse_patients_file(
patients_file_mouse, self.h2_database)
self.neoantigens = ModelConverter.parse_candidate_file(input_file)
self.neoantigens_mouse = ModelConverter.parse_candidate_file(
input_file)
def test_patients_csv_file2model_without_mhc1(self):
patients_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/patient.Pt29.without_mhc1.csv"
)
patients = ModelConverter.parse_patients_file(patients_file, self.hla_database)
self.assertIsNotNone(patients)
self.assertIsInstance(patients, list)
self.assertTrue(len(patients) == 1)
self.assertIsInstance(patients[0], Patient)
self.assertEqual(patients[0].identifier, "Pt29")
for m in patients[0].mhc1:
self.assertEqual(m.zygosity, Zygosity.LOSS)
self.assertEqual(3, len(patients[0].mhc2))
self.assertEqual(
9, len([a for m in patients[0].mhc2 for g in m.genes for a in g.alleles])
)
self.assertEqual(patients[0].is_rna_available, True)
def test_patients_csv_file2model_mouse(self):
patients_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/alleles.Pt29_mouse.csv"
)
patients = ModelConverter.parse_patients_file(patients_file, self.h2_database)
self.assertIsNotNone(patients)
self.assertIsInstance(patients, list)
self.assertTrue(len(patients) == 1)
self.assertIsInstance(patients[0], Patient)
self.assertEqual(patients[0].identifier, "Pt29")
self.assertEqual(3, len(patients[0].mhc1))
self.assertEqual(6, len([a for m in patients[0].mhc1 for a in m.alleles]))
self.assertEqual(2, len(patients[0].mhc2))
self.assertEqual(
3, len([a for m in patients[0].mhc2 for g in m.genes for a in g.alleles])
)
self.assertEqual(patients[0].is_rna_available, False)
def test_neofox_synthetic_data(self):
"""
this test just ensures that NeoFox does not crash with the synthetic data
"""
data = [
("resources/synthetic_data/neoantigens_1patients_10neoantigens.2.txt",
"resources/synthetic_data/patients_1patients_10neoantigens.2.txt"
),
("resources/synthetic_data/neoantigens_10patients_10neoantigens.0.txt",
"resources/synthetic_data/patients_10patients_10neoantigens.0.txt"
),
#("resources/synthetic_data/neoantigens_100patients_10neoantigens.2.txt",
# "resources/synthetic_data/patients_100patients_10neoantigens.2.txt"),
#("resources/synthetic_data/neoantigens_no_wt_1patients_10neoantigens.3.txt",
# "resources/synthetic_data/patients_no_wt_1patients_10neoantigens.3.txt"),
#("resources/synthetic_data/poltergeist_neoantigens.txt",
# "resources/synthetic_data/poltergeist_patients.txt")
("resources/synthetic_data/neoantigens_no_wt_10patients_10neoantigens.4.txt",
"resources/synthetic_data/patients_no_wt_10patients_10neoantigens.4.txt"
),
#("resources/synthetic_data/neoantigens_100patients_10neoantigens.4.txt",
# "resources/synthetic_data/patients_100patients_10neoantigens.4.txt"),
]
for n, p, in data:
input_file = pkg_resources.resource_filename(
neofox.tests.__name__, n)
data = pd.read_csv(input_file, sep="\t")
data = data.replace({np.nan: None})
neoantigens = ModelConverter._neoantigens_csv2objects(data)
patients_file = pkg_resources.resource_filename(
neofox.tests.__name__, p)
patients = ModelConverter.parse_patients_file(
patients_file, self.hla_database)
annotations = NeoFox(
neoantigens=neoantigens,
patients=patients,
num_cpus=4,
).get_annotations()
self.assertIsNotNone(annotations)
def test_no_expression_imputation(self):
input_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_candidate_file.txt"
)
patients_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_patient_file.txt"
)
patients = ModelConverter.parse_patients_file(patients_file, self.hla_database)
neoantigens = ModelConverter.parse_candidate_file(input_file)
neofox_runner = NeoFox(
neoantigens=neoantigens,
patients=patients,
reference_folder=FakeReferenceFolder(),
configuration=FakeDependenciesConfiguration(),
)
for neoantigen in neoantigens:
for neoantigen_imputed in neofox_runner.neoantigens:
if neoantigen.mutation.mutated_xmer == neoantigen_imputed.mutation.mutated_xmer:
self.assertEqual(
neoantigen.rna_expression, neoantigen_imputed.rna_expression
)
def test_with_expression_imputation(self):
input_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_candidate_file_Pty.txt"
)
neoantigens= ModelConverter.parse_candidate_file(input_file)
import copy
original_neoantigens = copy.deepcopy(neoantigens)
patients_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_patient_file.txt"
)
patients = ModelConverter.parse_patients_file(patients_file, self.hla_database)
neofox_runner = NeoFox(
neoantigens=neoantigens,
patients=patients,
reference_folder=FakeReferenceFolder(),
configuration=FakeDependenciesConfiguration(),
)
for neoantigen in original_neoantigens:
for neoantigen_imputed in neofox_runner.neoantigens:
self.assertFalse(
neoantigen.rna_expression == neoantigen_imputed.rna_expression
)
def test_patients_csv_file2model3(self):
patients_file = pkg_resources.resource_filename(
neofox.tests.__name__, "resources/test_patient_info.txt"
)
patients = ModelConverter.parse_patients_file(patients_file, self.hla_database)
self.assertIsNotNone(patients)
self.assertIsInstance(patients, list)
self.assertTrue(len(patients) == 1)
self.assertIsInstance(patients[0], Patient)
self.assertEqual(patients[0].identifier, "Ptx")
self.assertEqual(3, len(patients[0].mhc1))
self.assertEqual(6, len([a for m in patients[0].mhc1 for a in m.alleles]))
self.assertEqual(3, len(patients[0].mhc2))
self.assertEqual(
10, len([a for m in patients[0].mhc2 for g in m.genes for a in g.alleles])
)
self.assertTrue(
"HLA-A*03:01" in [a.name for m in patients[0].mhc1 for a in m.alleles]
)
self.assertTrue(
"HLA-DQA1*04:01"
in [a.name for m in patients[0].mhc2 for g in m.genes for a in g.alleles]
)
self.assertTrue(patients[0].is_rna_available)