def reannotate_trials():
db = database.get_db()
trials = list(db['trial'].find())
# modify trials to be inserted in bulk later
events.trial_insert(trials)
# re-insert.
for trial in trials:
db['trial'].delete_one({'_id': trial['_id']})
db['trial'].insert_one(trial)
logging.info("DONE")
def reannotate_trials():
# connect to the database.
db = database.get_db()
# get all trials.
trials = list(db['trial'].find())
# call hooks.
events.trial_insert(trials)
# re-insert.
for trial in trials:
db['trial'].delete_one({'_id': trial['_id']})
db['trial'].insert_one(trial)
def test_trial_normalization(self):
db = get_db()
db['normalize'].insert(MAPPING)
# define test json
with open(os.path.join(YAML_DIR, "00-001.yml")) as fin:
test_json = yaml.load(fin)
# do the mapping.
UtilHooks.NormalizeHook.entry_insert("trial", [test_json])
trial_insert([test_json])
db['normalize'].drop()
def test_trial_normalization(self):
db = get_db()
db['normalize'].insert(MAPPING)
# define test json
with open(os.path.join(YAML_DIR, "00-001.yml")) as fin:
test_json = yaml.load(fin)
# do the mapping.
UtilHooks.NormalizeHook.entry_insert("trial", [test_json])
trial_insert([test_json])
db['normalize'].drop()
def reannotate_trials():
# connect to the database.
db = database.get_db()
# get all trials.
trials = list(db['trial'].find())
# call hooks.
events.trial_insert(trials)
# re-insert.
for trial in trials:
db['trial'].delete_one({'_id': trial['_id']})
db['trial'].insert_one(trial)
def test_updatedt_with_curatedt(self):
# set up
today = dt.datetime.now().strftime('%B %d, %Y')
old = 'October 27, 2016'
on_trial['_genomic'] = {'hugo_symbol': [{'value': 'TEST'}]}
on_trial['_clinical'] = {'disease_status': [{'value': ['TEST']}]}
# set both dates to OLD
on_trial['curated_on'] = old
on_trial['last_updated'] = old
assert on_trial['curated_on'] == old, self._debug(
on_trial, 'curated_on')
assert on_trial['last_updated'] == old, self._debug(
on_trial, 'last_updated')
# set "last_updated" to TODAY
trial = set_updated(on_trial)
assert trial['curated_on'] == old, self._debug(trial, 'curated_on')
assert trial['last_updated'] == today, self._debug(
trial, 'last_updated')
# check dates are preserved after trial_insert is called
trial['protocol_no'] = '01-001'
trial = trial_insert([trial])
trial = trial[0]
assert trial['curated_on'] == old, self._debug(trial, 'curated_on')
assert trial['last_updated'] == today, self._debug(
trial, 'last_updated')
# check the other way around
on_trial['curated_on'] = old
trial['last_updated'] = old
assert trial['curated_on'] == old, self._debug(trial, 'curated_on')
assert trial['last_updated'] == old, self._debug(trial, 'last_updated')
trial = set_curated(on_trial)
assert trial['curated_on'] == today, self._debug(trial, 'curated_on')
assert trial['last_updated'] == today, self._debug(
trial, 'last_updated')
trial = trial_insert([trial])
trial = trial[0]
assert trial['curated_on'] == today, self._debug(trial, 'curated_on')
assert trial['last_updated'] == today, self._debug(
trial, 'last_updated')
def test_trial_insert(self):
# loop over each trial.
executed_new_yaml = False
for trial_name in os.listdir(YAML_DIR):
# open the trial as json
trial_path = os.path.join(YAML_DIR, trial_name)
with open(trial_path, "rb") as fin:
# load yaml
data_json = yaml.load(fin.read())
# normalize.
UtilHooks.NormalizeHook.entry_insert('trial', [data_json])
# insert it.
trial_insert([data_json])
# check these fields are present.
assert '_summary' in data_json
# trial 15-153 contains hormone receptor status and wildtype hugo symbol fields
if trial_name == '00-002.yml':
assert 'PIK3CA' in data_json['_summary']["genes"]
assert 'wt PIK3CA' in data_json['_summary']["genes"]
assert 'Invasive Breast Carcinoma ER+/HER2-' in data_json[
'_summary']['tumor_types']
assert len(data_json['_summary']['tumor_types']) == 1
assert set(data_json['_summary']['disease_status']) == {
'Metastatic', 'Advanced'
}
executed_new_yaml = True
if trial_name == '00-001.yml':
for diag in data_json['_summary']['tumor_types']:
assert diag[0] != '!'
assert executed_new_yaml, 'Please add trial 00-002 to the tests/data/yaml directory so that' \
'hormone receptor status and wildtype hugo symbol functionality can be' \
'adequately tested.'
def test_updatedt_with_curatedt(self):
# set up
today = dt.datetime.now().strftime('%B %d, %Y')
old = 'October 27, 2016'
on_trial['_genomic'] = {'hugo_symbol': [{'value': 'TEST'}]}
on_trial['_clinical'] = {'disease_status': [{'value': ['TEST']}]}
# set both dates to OLD
on_trial['curated_on'] = old
on_trial['last_updated'] = old
assert on_trial['curated_on'] == old, self._debug(on_trial, 'curated_on')
assert on_trial['last_updated'] == old, self._debug(on_trial, 'last_updated')
# set "last_updated" to TODAY
trial = set_updated(on_trial)
assert trial['curated_on'] == old, self._debug(trial, 'curated_on')
assert trial['last_updated'] == today, self._debug(trial, 'last_updated')
# check dates are preserved after trial_insert is called
trial = trial_insert([trial])
trial = trial[0]
assert trial['curated_on'] == old, self._debug(trial, 'curated_on')
assert trial['last_updated'] == today, self._debug(trial, 'last_updated')
# check the other way around
on_trial['curated_on'] = old
trial['last_updated'] = old
assert trial['curated_on'] == old, self._debug(trial, 'curated_on')
assert trial['last_updated'] == old, self._debug(trial, 'last_updated')
trial = set_curated(on_trial)
assert trial['curated_on'] == today, self._debug(trial, 'curated_on')
assert trial['last_updated'] == today, self._debug(trial, 'last_updated')
trial = trial_insert([trial])
trial = trial[0]
assert trial['curated_on'] == today, self._debug(trial, 'curated_on')
assert trial['last_updated'] == today, self._debug(trial, 'last_updated')
def test_trial_insert(self):
# loop over each trial.
executed_new_yaml = False
for trial_name in os.listdir(YAML_DIR):
# open the trial as json
trial_path = os.path.join(YAML_DIR, trial_name)
with open(trial_path, "rb") as fin:
# load yaml
data_json = yaml.load(fin.read())
# normalize.
UtilHooks.NormalizeHook.entry_insert('trial', [data_json])
# insert it.
trial_insert([data_json])
# check these fields are present.
assert '_summary' in data_json
# trial 15-153 contains hormone receptor status and wildtype hugo symbol fields
if trial_name == '00-002.yml':
assert 'PIK3CA' in data_json['_summary']["genes"]
assert 'wt PIK3CA' in data_json['_summary']["genes"]
assert 'Invasive Breast Carcinoma ER+/HER2-' in data_json['_summary']['tumor_types']
assert len(data_json['_summary']['tumor_types']) == 1
assert data_json['_summary']['disease_status'] == ['Metastatic', 'Advanced']
executed_new_yaml = True
if trial_name == '00-001.yml':
for diag in data_json['_summary']['tumor_types']:
assert diag[0] != '!'
assert executed_new_yaml, 'Please add trial 00-002 to the tests/data/yaml directory so that' \
'hormone receptor status and wildtype hugo symbol functionality can be' \
'adequately tested.'
