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)
column_name = feature_fields[0]
table_field, value_type = cls.get_table_field_and_value_type(column_name)
if table_field is None:
raise FeatureNotFoundException(feature_id)
return cls(table_field, value_type)
def from_feature_id(cls, feature_id):
feature_fields = cls.regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
logging.debug(feature_fields)
value_field, chromosome, start, end = feature_fields[0]
valid_chr_set = frozenset([str(x)
for x in range(1, 24)] + ['X', 'Y', 'M'])
if chromosome not in valid_chr_set:
raise FeatureNotFoundException(feature_id)
return cls(value_field, chromosome, start, end)
def parse_internal_feature_id(self, feature_id):
# TODO Better input validation
feature_type, gene_label, value_field = feature_id.split(':')
if value_field not in VALUES:
raise FeatureNotFoundException(feature_id)
self.feature_type = feature_type
self.gene_label = gene_label
self.value_field = value_field
self.table_info = get_table_info()
if self.table_info is None:
raise FeatureNotFoundException(feature_id)
self.table_name = self.table_info['name']
def from_feature_id(cls, feature_id, project_id=None):
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("^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 = feature_fields[0]
return cls(gene_label, protein_name)
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 = cls.get_feature_type_string(feature_id)
if feature_type is None:
logging.debug("FeatureProviderFactory.from_feature_id: invalid feature ID: " + str(feature_id))
raise FeatureNotFoundException(feature_id)
if feature_type == CLINICAL_FEATURE_TYPE:
return ClinicalFeatureProvider
elif feature_type == GEXP_FEATURE_TYPE:
return GEXPFeatureProvider
elif feature_type == METH_FEATURE_TYPE:
return METHFeatureProvider
elif feature_type == CNVR_FEATURE_TYPE:
return CNVRFeatureProvider
elif feature_type == RPPA_FEATURE_TYPE:
return RPPAFeatureProvider
elif feature_type == MIRN_FEATURE_TYPE:
return MIRNFeatureProvider
elif feature_type == GNAB_FEATURE_TYPE:
return GNABFeatureProvider
elif feature_type == USER_FEATURE_TYPE:
return UserFeatureProvider
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, table_id = feature_fields[0]
value_field = cls.get_table_info(table_id)['value_field']
return cls(gene_label, value_field, table_id)
def parse_internal_feature_id(self, feature_id):
self.feature_def = GNABFeatureDef.from_feature_id(feature_id)
self.table_info = self.get_table_info()
if self.table_info is None:
raise FeatureNotFoundException(feature_id)
self.table_name = self.table_info['name']
def from_feature_id(cls, feature_id):
feature_fields = cls.regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
mirna_name, table_id = feature_fields[0]
table_info = cls.get_table_info(table_id)
platform = table_info['platform']
value_field = table_info['value_field']
return cls(mirna_name, platform, value_field, table_id)
def from_feature_id(cls, feature_id):
# Example ID: METH:cg08246323:HumanMethylation450:methylation_chr16
regex = re_compile("^METH:"
# TODO better validation for probe name
"([a-zA-Z0-9_.\-]+):"
# platform
"(HumanMethylation27|HumanMethylation450):"
# validate outside - chromosome 1-23, X, Y, M
"methylation_chr(\d|\d\d|X|Y|M)$")
feature_fields = regex.findall(feature_id)
if len(feature_fields) == 0:
raise FeatureNotFoundException(feature_id)
probe, platform, chromosome = 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(probe, platform, chromosome)
def parse_internal_feature_id(self, feature_id):
# TODO better feature ID input validation
feature_type, gene_label, table_id = feature_id.split(':')
self.feature_type = feature_type
self.gene_label = gene_label
self.table_id = table_id
self.table_info = get_table_info(table_id)
if self.table_info is None:
raise FeatureNotFoundException(feature_id)
self.table_name = self.table_info['table_id']
self.value_field = self.table_info['value_field']