def setUpClass(cls):
# Clean up before running each test, in case quited test before tearDown() was run, eg when debugging
cls.__cleanup()
# Use test YAML configuration file
Configuration(cls.__test_configuration_file)
# Delete test database if it exists
if os.path.exists(Configuration().database_location):
os.remove(Configuration().database_location)
# Create the database
create_database_controller.run()
# Populate the database
download_cancer_genomics_controller.run()
# Pre-prepare the downloaded data by converting it into a form that can be used for machine learning
prepare_machine_learning_data_controller.run()
def test_tables_have_been_created(self):
# Create the database
create_database_controller.run()
# Connect to the database, and get all of the table names into a list
db = sqlite3.connect(Configuration().database_location)
cursor = db.cursor()
results = cursor.execute(
"SELECT name FROM sqlite_master WHERE type='table';").fetchall()
# At least 5 tables should have been created (more detailed tests are in a separate test class)
self.assertGreaterEqual(len(results), 5)
def run():
logger = Configuration().get_logger(__name__)
try:
logger.info("Starting predicting_cancer_variants")
create_database_controller.run()
download_cancer_genomics_controller.run()
prepare_machine_learning_data_controller.run()
logger.info("predicting_cancer_variants complete")
except Exception as e:
message = "predicting_cancer_variants has failed"
logger.exception(message, exc_info=e)
raise SystemExit(message)
def run():
logger = Configuration().get_logger(__name__)
try:
logger.info("Preparing machine learning data")
StagingTable().prepare()
PatientTable().prepare()
NetDataTable().prepare()
logger.info("Preparation of machine learning data complete")
except Exception as e:
message = "Failed to prepare machine learning data"
logger.exception(message, exc_info=e)
raise SystemExit(message)
def test_error_is_raised_if_config_file_does_not_exist(self):
non_existent_configuration_file = r"c:\this\file\does\not-exist.yaml"
c = Configuration(non_existent_configuration_file)
try:
# this should raise an error because the config file does not exist
c._Configuration__get_configuration_filename()
# should not reach here
self.fail("An exception was not raised")
except FileNotFoundError as e:
# expected to reach here
self.assertEqual(
str(e), "This configuration file does not exist: '{}'".format(
non_existent_configuration_file))
pass
except:
self.fail("The wrong exception was raised")
def run():
logger = Configuration().get_logger(__name__)
try:
logger.info("Creating tables")
StudiesTable().create()
CancerTable().create()
MolecularTable().create()
MutationTable().create()
GeneTable().create()
logger.info("Table creation complete")
except Exception as e:
message = "Failed to create tables"
logger.exception(message, exc_info=e)
raise SystemExit(message)
def run():
logger = Configuration().get_logger(__name__)
try:
logger.info(
"Downloading and storing data from large-scale online Cancer Genomics database cBioPortal.org"
)
StudiesDownloader().download_and_save_all_data()
GeneDownloader().download_and_save_all_data()
CancerDownloader().download_and_save_all_data()
MolecularDownloader().download_and_save_data_every_large_study()
MutationDownloader().download_and_save_data_every_large_study()
logger.info("Downloads complete")
except Exception as e:
message = "Failed to download and store data"
logger.exception(message, exc_info=e)
raise SystemExit(message)
def __init__(self,
cancer_ids=[],
manually_test_model=False,
show_confusion=True,
cross_validate=True,
compare_with_dummy=True,
precision_and_recall_verbosity=True,
show_precision_and_recall=True):
self._configuration = Configuration()
self.__logger = self._configuration.get_logger(__name__)
self.__retriever = DataRetriever()
self.df = self.__retriever.fetch()
self.cancer_ids = cancer_ids
self.show_confusion = show_confusion
self.cross_validate = cross_validate
self.manually_test_model = manually_test_model
self.compare_with_dummy = compare_with_dummy
self.verbose_level_precision_and_recall = precision_and_recall_verbosity
self.show_precision_and_recall_graph = show_precision_and_recall
def __init__(self):
self._configuration = Configuration()
self.__logger = self._configuration.get_logger(__name__)
def test_can_fetch_study_url(self):
c = Configuration()
self.assertIn("www.cbioportal.org", c.study_url)
def test_can_fetch_cancer_table(self):
c = Configuration(
r"Test Configuration Files\test_configuration_file_absolute_path.yaml"
)
self.assertEqual("test_cancer_study", c.database_table_study)
def test_can_fetch_database_location_with_absolute_path(self):
c = Configuration(
r"Test Configuration Files\test_configuration_file_absolute_path.yaml"
)
self.assertEqual(r"c:\a-test\location\file.yaml", c.database_location)
def test_default_config_file_is_correct(self):
c = Configuration()
self.assertIn(r'Configuration Files\predicting_cancer_variants.yaml',
c._Configuration__get_default_configuration_filename())
def test_application_root_is_ok(self):
c = Configuration()
self.assertIn('predicting_cancer_variants', c.application_root)
def test_database_file_exists(self):
# Database should now exist
self.assertTrue(os.path.exists(Configuration().database_location))
def test_error_is_not_raised_if_config_file_does_exist(self):
c = Configuration(
r"Test Configuration Files\test_configuration_file_absolute_path.yaml"
)
self.assertIn(r'test_configuration_file_absolute_path.yaml',
c._Configuration__get_configuration_filename())
def __init__(self):
self.__database_helper = DatabaseHelper()
self.__configuration = Configuration()
self.__logger = self.__configuration.get_logger(__name__)