def user_feature_handler(feature_id, cohort_id_array):
include_tcga = False
user_studies = ()
for cohort_id in cohort_id_array:
try:
db = sql_connection()
cursor = db.cursor(MySQLdb.cursors.DictCursor)
cursor.execute("SELECT project_id FROM cohorts_samples WHERE cohort_id = %s GROUP BY project_id", (cohort_id,))
for row in cursor.fetchall():
if row['project_id'] is None:
include_tcga = True
else:
user_studies += (row['project_id'],)
except Exception as e:
if db: db.close()
if cursor: cursor.close()
raise e
user_feature_id = None
if feature_id.startswith('USER:'******'t include TCGA
include_tcga = False
return {
'converted_feature_id': feature_id,
'include_tcga': include_tcga,
'user_studies': user_studies,
'user_feature_id': user_feature_id
}
def get_feature_vector(feature_id, cohort_id_array):
include_tcga = False
user_studies = ()
for cohort_id in cohort_id_array:
try:
db = sql_connection()
cursor = db.cursor(MySQLdb.cursors.DictCursor)
cursor.execute("SELECT project_id FROM cohorts_samples WHERE cohort_id = %s GROUP BY project_id", (cohort_id,))
for row in cursor.fetchall():
if row['project_id'] is None:
include_tcga = True
else:
user_studies += (row['project_id'],)
except Exception as e:
if db: db.close()
if cursor: cursor.close()
raise e
# ex: feature_id 'CLIN:Disease_Code'
user_feature_id = None
if feature_id.startswith('USER:'******'t include TCGA
include_tcga = False
items = []
type = None
result = []
cohort_settings = settings.GET_BQ_COHORT_SETTINGS()
if include_tcga:
provider = FeatureProviderFactory.from_feature_id(feature_id)
result = provider.get_data(cohort_id_array, cohort_settings.dataset_id, cohort_settings.table_id)
# ex: result[0]
# {'aliquot_id': None, 'case_id': u'TCGA-BH-A0B1', 'sample_id': u'TCGA-BH-A0B1-10A', 'value': u'BRCA'}
for data_point in result:
data_item = {key: data_point[key] for key in ['case_id', 'sample_id', 'aliquot_id']}
value = provider.process_data_point(data_point)
# TODO refactor missing value logic
if value is None:
value = 'NA'
data_item['value'] = value
items.append(data_item)
type = provider.get_value_type()
if len(user_studies) > 0:
# Query User Data
user_provider = UserFeatureProvider(feature_id, user_feature_id=user_feature_id)
user_result = user_provider.get_data(cohort_id_array, cohort_settings.dataset_id, cohort_settings.table_id)
result.extend(user_result)
for data_point in user_result:
data_item = {key: data_point[key] for key in ['case_id', 'sample_id', 'aliquot_id']}
value = provider.process_data_point(data_point)
# TODO refactor missing value logic
if value is None:
value = 'NA'
data_item['value'] = value
items.append(data_item)
if not type:
type = user_provider.get_value_type()
return type, items