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.
entry_insert("trial", [test_json])
build_trial_elasticsearch_fields([test_json])
db['normalize'].drop()
def test_trial_validation(self):
# define test json
with open(os.path.join(YAML_DIR, "00-003.yml")) as fin:
test_json = yaml.load(fin)
# post and it should fail.
r, status_code = self.post("trial", test_json)
assert status_code == 422, r
# add values to database.
self.db['normalize'].drop()
mapping = copy.deepcopy(self.mapping)
mapping['values']['oncotree_primary_diagnosis'][
"Diffuse Large B-Cell Lymphoma"] = "Diffuse Large B-Cell Lymphoma"
self.db['normalize'].insert(mapping)
# post and it should pass
build_trial_elasticsearch_fields([test_json])
assert test_json['treatment_list']['step'][1]['arm'][1]['match'][0][
'and'][1]['clinical'][
'oncotree_primary_diagnosis'] == '!Diffuse Large B-Cell Lymphoma'
assert test_json['status'] == 'OPEN TO ACCRUAL'
suggestor_options = [
'cancer_type_suggest', 'hugo_symbol_suggest', 'variant_suggest',
'wildtype_suggest', 'cnv_suggest', 'sv_suggest',
'protocol_no_suggest', 'disease_center_suggest',
'disease_status_suggest', 'drug_suggest', 'investigator_suggest',
'mmr_status_suggest', 'nct_number_suggest'
]
assert sorted(
test_json['_suggest'].keys()) == sorted(suggestor_options)
genes = test_json['_suggest']['hugo_symbol_suggest']['input']
assert isinstance(genes, list)
assert 'MYC' in genes
assert len(genes) == 1
assert test_json['_suggest']['cancer_type_suggest'] == []
def test_set_update(self):
trial = set_updated(self.trial)
trial = self._prep_trial(trial)
assert 'last_updated' in trial, trial
build_trial_elasticsearch_fields([trial])
# send again
trial = set_updated(trial)
build_trial_elasticsearch_fields([trial])
assert 'last_updated' in trial, trial
assert 'curated_on' in trial and trial['curated_on'] == '', trial
for field in self.trial_status_fields:
assert field in trial['_summary'], '%s\n\n %s' % (
trial['_summary'], field)
# curate
trial = set_curated(trial)
assert 'last_updated' in trial, trial
assert 'curated_on' in trial and trial['curated_on'], trial
for field in self.trial_status_fields:
assert field in trial['_summary'], '%s\n\n%s' % (trial['_summary'],
field)
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'
build_trial_elasticsearch_fields([trial])
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')
build_trial_elasticsearch_fields([trial])
assert trial['curated_on'] == today, self._debug(trial, 'curated_on')
assert trial['last_updated'] == today, self._debug(trial, 'last_updated')
def test_trial_reinsert(self):
# remove extra fields.
fields = ['_summary']
trial = yaml_schema.yaml_test_json.copy()
for key in fields:
if key in trial:
del trial[key]
# insert it.
self.db['trial'].insert(trial)
# check we don't have fields.
for trial in self.db['trial'].find():
for x in fields:
assert x not in trial
# re-annotate it.
build_trial_elasticsearch_fields([trial])
# assert we have the fields in the output.
for x in fields:
assert x in trial
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.
entry_insert('trial', [data_json])
# insert it.
build_trial_elasticsearch_fields([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.'