def test_affinity_threshold(self):
result = TcellPrediction(affinity_threshold=1)._calculate_tcell_predictor_score(
gene="BRCA2",
substitution="CCCCVCCCC",
epitope="CCCCCCCCC",
score=5
)
self.assertIsNone(result)
def test_existing_gene(self):
result = TcellPrediction(affinity_threshold=10)._calculate_tcell_predictor_score(
gene="BRCA2",
substitution="CCCCVCCCC",
epitope="CCCCCCCCC",
score=5
)
self.assertAlmostEqual(0.2453409331088489, float(result))
def test_non_existing_gene(self):
result = TcellPrediction(affinity_threshold=10)._calculate_tcell_predictor_score(
gene="BLAH",
substitution="blaaaah",
epitope="BLAHBLAH",
score=5
)
self.assertEqual(None, result)
def test_existing_gene_with_too_long_epitope(self):
result = TcellPrediction(affinity_threshold=10)._calculate_tcell_predictor_score(
gene="BRCA2", substitution="C", epitope="CCCCCCCCCC", score=5
)
self.assertEqual(None, result)
def setUp(self) -> None:
self.tcell_predictor = TcellPrediction()
def __init__(self,
neoantigens: List[Neoantigen],
patients: List[Patient],
num_cpus: int = 1,
patient_id: str = None,
work_folder=None,
output_prefix=None,
reference_folder: ReferenceFolder = None,
configuration: DependenciesConfiguration = None,
verbose=True,
configuration_file=None,
affinity_threshold=AFFINITY_THRESHOLD_DEFAULT):
self.affinity_threshold = affinity_threshold
if configuration_file:
dotenv.load_dotenv(configuration_file, override=True)
# initialise logs
self.log_file_name = self._get_log_file_name(output_prefix,
work_folder)
self._initialise_logs(self.log_file_name, verbose)
# intialize references folder and configuration
# NOTE: uses the reference folder and config passed as a parameter if exists, this is here to make it
# testable with fake objects
self.reference_folder = (reference_folder if reference_folder else
ReferenceFolder(verbose=verbose))
# NOTE: makes this call to force the loading of the available alleles here
self.reference_folder.get_available_alleles()
self.configuration = (configuration if configuration else
DependenciesConfiguration())
self.tcell_predictor = TcellPrediction(
affinity_threshold=self.affinity_threshold)
self.self_similarity = SelfSimilarityCalculator()
self.num_cpus = num_cpus
if (neoantigens is None or len(neoantigens) == 0 or patients is None
or len(patients) == 0):
raise NeofoxConfigurationException(
"Missing input data to run Neofox")
# validates neoantigens
self.neoantigens = neoantigens
for n in self.neoantigens:
if n.patient_identifier is None:
n.patient_identifier = patient_id
# NOTE: the position of the mutations is not expected from the user and if provide the value is ignored
n.mutation.position = EpitopeHelper.mut_position_xmer_seq(
mutation=n.mutation)
ModelValidator.validate_neoantigen(n)
# validates patients
self.patients = {}
for patient in patients:
ModelValidator.validate_patient(
patient, organism=self.reference_folder.organism)
self.patients[patient.identifier] = patient
self._validate_input_data()
# retrieve from the data, if RNA-seq was available
# add this information to patient model
expression_per_patient = {
self.patients[patient].identifier: []
for patient in self.patients
}
for neoantigen in self.neoantigens:
expression_per_patient[neoantigen.patient_identifier].append(
neoantigen.rna_expression)
for patient in self.patients:
self.patients[patient].is_rna_available = all(
e is not None for e in expression_per_patient[
self.patients[patient].identifier])
# only performs the expression imputation for humans
if self.reference_folder.organism == ORGANISM_HOMO_SAPIENS:
# impute expresssion from TCGA, ONLY if isRNAavailable = False for given patient,
# otherwise original values is reported
# NOTE: this must happen after validation to avoid uncaptured errors due to missing patients
# NOTE: add gene expression to neoantigen candidate model
self.neoantigens = self._conditional_expression_imputation()
logger.info("Data loaded")