def get_provider_class_from_feature_id(cls, feature_id):
"""
Args:
feature_id: Feature identifier
Returns:
Feature data provider class for the datatype defined in the
feature identifier.
Raises:
FeatureNotFoundException: If the datatype part of the feature
identifier is unknown.
"""
feature_type_prefix = cls.get_feature_type_string(feature_id)
if feature_type_prefix is None:
logger.error(
"FeatureProviderFactory.from_feature_id: Unknown type: " +
str(feature_id))
raise FeatureNotFoundException(feature_id)
if feature_type_prefix.lower() not in FEATURE_ID_TO_TYPE_MAP:
logger.error(
"FeatureProviderFactory.from_feature_id: invalid feature ID: "
+ str(feature_id))
raise FeatureNotFoundException(feature_id)
feature_type = FEATURE_ID_TO_TYPE_MAP[feature_type_prefix.lower()]
return FeatureDataTypeHelper.get_feature_data_provider_from_data_type(
feature_type)
def from_user_feature_id(cls, feature_id):
logging.debug("UserFeatureDef.from_user_feature_id {0}".format(
str([feature_id])))
# ID breakdown: project ID:Feature ID
# Example ID: USER:1:6
regex = re_compile("^USER:"******"([0-9]+):"
# Feature ID
"([0-9]+)$")
feature_fields = regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
project_id, user_feature_id = feature_fields[0]
bq_id = None
shared_id = None
is_numeric = False
try:
db = get_sql_connection()
cursor = db.cursor(MySQLdb.cursors.DictCursor)
cursor.execute(
"""
SELECT feature_name, bq_map_id, shared_map_id, is_numeric
FROM projects_user_feature_definitions
WHERE id = %s
""", (user_feature_id, ))
for row in cursor.fetchall():
if row['shared_map_id']:
shared_id = row['shared_map_id']
bq_id = row["bq_map_id"]
is_numeric = row['is_numeric'] == 1
cursor.close()
db.close()
except Exception as e:
if db: db.close()
if cursor: cursor.close()
raise e
if shared_id is not None:
return cls.from_feature_id(bq_id, project_id)
if bq_id is None:
raise FeatureNotFoundException(feature_id)
# Else we're querying a very specific feature from a specific project
bq_table, column_name, symbol = cls.get_table_and_field(bq_id)
if bq_table is None or column_name is None:
raise FeatureNotFoundException(feature_id)
logging.debug("{0} {1} {2}".format(bq_table, column_name, symbol))
filters = None
if symbol is not None:
filters = {'Symbol': symbol}
return [cls(bq_table, column_name, project_id, is_numeric, filters)]
def from_feature_id(cls, feature_id):
feature_fields = cls.regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
value_field, chromosome, start, end, internal_table_id = feature_fields[0]
valid_chr_set = frozenset([str(x) for x in xrange(1, 24)] + ['X', 'Y', 'M'])
if chromosome not in valid_chr_set:
raise FeatureNotFoundException(feature_id)
return cls(value_field, chromosome, start, end, internal_table_id)
def from_feature_id(cls, feature_id, project_id=None):
"""
This is the method used when the user feature maps to feature in an existing project (i.e. there is a
shared_id in the projects_user_feature_definitions entry).
It returns a *LIST* of UserFeatureDefs
"""
logging.debug("UserFeatureDef.from_feature_id: {0}".format(
str([feature_id, project_id])))
if feature_id is None:
raise FeatureNotFoundException(feature_id)
# ID breakdown: project ID:Feature ID
# Example ID: USER:1:6
regex = re_compile("(?:^v2:)?USER:"******"([0-9]+):"
# Feature ID
"([0-9]+)$")
feature_fields = regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
project_id, user_feature_id = feature_fields[0]
try:
db = get_sql_connection()
cursor = db.cursor(MySQLdb.cursors.DictCursor)
cursor.execute(
"""
SELECT bq_map_id, project_id, is_numeric
FROM projects_user_feature_definitions
WHERE id = %s
""", (user_feature_id, ))
results = []
for row in cursor.fetchall():
bq_table, column_name, symbol = cls.get_table_and_field(
row['bq_map_id'])
filters = None
if symbol is not None:
filters = {'Symbol': symbol}
results.append(
cls(bq_table, column_name, row['project_id'],
row['is_numeric'] == 1, filters))
cursor.close()
db.close()
return results
except Exception as e:
if db: db.close()
if cursor: cursor.close()
raise e
def from_feature_id(cls, feature_id):
feature_fields = cls.regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
gene_label, protein_name, internal_table_id = feature_fields[0]
return cls(gene_label, protein_name, internal_table_id)
def from_feature_id(cls, feature_id):
feature_fields = cls.regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
gene_label, genomic_build = feature_fields[0]
return cls(gene_label, genomic_build)
def from_feature_id(cls, feature_id):
feature_fields = cls.regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
column_name = feature_fields[0]
# Check if the column exists in any configured tables.
# If matching tables are found, then check that the value type
# of the column in the found tables is the same.
found_tables = []
for table_config in cls.config_instance.data_table_list:
schema = TABLE_TO_SCHEMA_MAP[table_config.table_id]
for field_item in schema:
if field_item['name'] == column_name:
# Capture the type of the field in this table
found_tables.append((table_config, field_item['type']))
break
if len(found_tables) == 0:
raise InvalidClinicalFeatureIDException(
feature_id, "No tables found for column name")
data_type_set = set([table[1] for table in found_tables])
if len(data_type_set) != 1:
raise InvalidClinicalFeatureIDException(
feature_id, "Data types of found tables do not match")
value_type = BigQuerySchemaToValueTypeConverter.get_value_type(
list(data_type_set)[0])
return cls(column_name, value_type)
def from_feature_id(cls, feature_id):
config_instance = METHDataSourceConfig.from_dict(BIGQUERY_CONFIG)
# Example ID: METH:cg08246323:HumanMethylation450:hg19_chr16
regex = re_compile("^v2:METH:"
# TODO better validation for probe name
"([a-zA-Z0-9_.\-]+):"
# platform
"(HumanMethylation27|HumanMethylation450):"
# internal table ID: 'hg19_chr1', 'hg19_chrx', etc
"(" + "|".join([
table.internal_table_id
for table in config_instance.data_table_list
]) + ")$")
feature_fields = regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
probe, platform, internal_table_id = feature_fields[0]
return cls(probe, platform, internal_table_id)
def from_user_feature_id(cls, feature_id):
"""
This returns a *LIST* of one UserFeatureDef, unless it maps to a standard project ID. Then
the list may have multiple entries.
"""
logging.debug("UserFeatureDef.from_user_feature_id {0}".format(
str([feature_id])))
# ID breakdown: project ID:Feature ID
# Example ID: USER:1:6
regex = re_compile("(?:^v2:)?USER:"******"([0-9]+):"
# Feature ID
"([0-9]+)$")
feature_fields = regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
project_id, user_feature_id = feature_fields[0]
bq_id = None
shared_id = None
is_numeric = False
try:
db = get_sql_connection()
cursor = db.cursor(MySQLdb.cursors.DictCursor)
cursor.execute(
"""
SELECT feature_name, bq_map_id, shared_map_id, is_numeric
FROM projects_user_feature_definitions
WHERE id = %s
""", (user_feature_id, ))
for row in cursor.fetchall():
if row['shared_map_id']:
shared_id = row['shared_map_id']
bq_id = row["bq_map_id"]
is_numeric = row['is_numeric'] == 1
cursor.close()
db.close()
except Exception as e:
if db: db.close()
if cursor: cursor.close()
raise e
# The feature we want is "shared", i.e. defined in a project we are extended from. So we delegate the
# task to the from_feature_id() function:
if shared_id is not None:
return cls.from_feature_id(bq_id, project_id)
if bq_id is None:
raise FeatureNotFoundException(feature_id)
# Else we're querying a very specific feature from a specific project
bq_table, column_name, symbol = cls.get_table_and_field(bq_id)
if bq_table is None or column_name is None:
raise FeatureNotFoundException(feature_id)
logging.debug("{0} {1} {2}".format(bq_table, column_name, symbol))
filters = None
if symbol is not None:
filters = {'Symbol': symbol}
return [cls(bq_table, column_name, project_id, is_numeric, filters)]
