def set_verbose(self, value):
if not isinstance(value, bool):
raise biomart.BiomartException("verbose must be set to a boolean value")
setattr(self, '_verbose', value)
# propagate verbose state to databases and datasets objects
for database in self._databases.values():
database.verbose = True
for dataset in self._datasets.values():
dataset.verbose = True
def __init__(self, *args, **kwargs):
server = kwargs.get('server', None)
if not server:
url = args[0]
server = biomart.BiomartServer(url=url, **kwargs)
self.server = server
self.name = kwargs.get('name', None)
if not self.name:
raise biomart.BiomartException(
"[BiomartDatabase] 'name' is required")
self.display_name = kwargs.get('display_name', self.name)
self.virtual_schema = kwargs.get('virtual_schema', 'default')
self.verbose = kwargs.get('verbose', False)
self._datasets = {}
def __init__(self, *args, **kwargs):
# dataset specific attributes
self.name = kwargs.get('name', None)
if not self.name:
raise biomart.BiomartException("[BiomartDataset] 'name' is required")
self.display_name = kwargs.get('display_name', self.name)
self.interface = kwargs.get('interface', 'default')
self.verbose = kwargs.get('verbose', False)
# get related biomart server
server = kwargs.get('server', None)
if not server:
url = args[0]
server = biomart.BiomartServer(url = url, **kwargs)
self.server = server
# get related biomart database
self.database = kwargs.get('database', None)
self._filters = {}
self._attribute_pages = {}
def search(self, params={}, header=0, count=False, formatter='TSV'):
if not isinstance(params, dict):
raise biomart.BiomartException("'params' argument must be a dict")
if self.verbose:
print("[BiomartDataset:'%s'] Searching using following params:" % self.name)
pprint.pprint(params)
# read filters and attributes from params
filters = params.get('filters', {})
attributes = params.get('attributes', [])
# check filters
for filter_name, filter_value in filters.items():
dataset_filter = self.filters.get(filter_name, None)
if not dataset_filter:
if self.verbose:
self.show_filters()
raise biomart.BiomartException("The filter '%s' does not exist." % filter_name)
accepted_values = dataset_filter.accepted_values
if len(accepted_values) > 0:
incorrect_value = None
if (isinstance(filter_value, list) or isinstance(filter_value, tuple)) and dataset_filter.filter_type == 'list':
incorrect_value = filter(lambda v: v not in accepted_values, filter_value)
elif filter_value not in accepted_values:
incorrect_value = filter_value
if incorrect_value:
error_msg = "the value(s) '%s' for filter '%s' cannot be used." % (incorrect_value, filter_name)
error_msg += " Use values from: [%s]" % ", ".join(map(str, accepted_values))
raise biomart.BiomartException(error_msg)
# check attributes unless we're only counting
if not count:
# discover attributes and pages
self.fetch_attributes()
# no attributes given, use default attributes
if not attributes and self._attribute_pages:
# get default attribute page
page = next(filter(lambda attr_page: attr_page.is_default, self._attribute_pages.values()))
# get default attributes from page
attributes = [a.name for a in page.attributes.values() if a.is_default]
# there is no default attributes, get all attributes from page
if not attributes:
attributes = [a.name for a in page.attributes.values()]
# if no default attributes have been defined, raise an exception
if not attributes:
raise biomart.BiomartException("at least one attribute is required, none given")
for attribute_name in attributes:
found = False
for page in self._attribute_pages.values():
if attribute_name in page.attributes.keys():
found = True
break
if not found:
if self.verbose:
self.show_attributes()
raise biomart.BiomartException("The attribute '%s' does not exist." % attribute_name)
# guess the attribute page and check if all attributes belong to it.
guessed_page = None
for tested_page in self._attribute_pages.values():
if set(attributes).issubset(tested_page.attributes.keys()):
guessed_page = tested_page
break
if guessed_page is None:
# selected attributes must belong to the same attribute page.
if self.verbose:
self.show_attributes()
raise biomart.BiomartException("You must use attributes that belong to the same attribute page.")
# filters and attributes looks ok, start building the XML query
root = Element('Query')
root.attrib.update({
'virtualSchemaName': self.database.virtual_schema,
'formatter': 'TSV',
'header': str(header),
'uniqueRows': '1',
'datasetConfigVersion': '0.6',
'count': count is True and '1' or ''
})
dataset = SubElement(root, "Dataset")
dataset.attrib.update({
'name': self.name,
'interface': self.interface
})
# Add filters to the XML query
for filter_name, filter_value in filters.items():
dataset_filter = self.filters[filter_name]
filter_elem = SubElement(dataset, "Filter")
filter_elem.set('name', filter_name)
if 'boolean_list' == dataset_filter.filter_type:
if filter_value is True or filter_value.lower() in ('included', 'only'):
filter_elem.set('excluded', '0')
elif filter_value is False or filter_value.lower() == 'excluded':
filter_elem.set('excluded', '1')
else:
if isinstance(filter_value, list) or isinstance(filter_value, tuple):
filter_value = ",".join(map(str, filter_value))
filter_elem.set('value', str(filter_value))
# Add attributes to the XML query, unless we're only counting
if not count:
for attribute_name in attributes:
attribute_elem = SubElement(dataset, "Attribute")
attribute_elem.set('name', str(attribute_name))
if self.verbose:
print("[BiomartDataset] search query:\n%s" % tostring(root))
return self.server.get_request(query = tostring(root))
def search(self, params = {}, header = 0, count = False):
if not isinstance(params, dict):
raise biomart.BiomartException("'params' argument must be a dict")
if self.verbose:
print("[BiomartDataset:'%s'] Searching using following params:" % self.name)
pprint.pprint(params)
# read filters and attributes from params
filters = params.get('filters', {})
attributes = params.get('attributes', [])
# check filters
for filter_name, filter_value in filters.items():
dataset_filter = self.filters.get(filter_name, None)
if not dataset_filter:
self.show_filters()
raise biomart.BiomartException("The filter '%s' does not exist." % filter_name)
if len(dataset_filter.accepted_values) > 0 and filter_value not in dataset_filter.accepted_values:
error_msg = "The value '%s' for filter '%s' cannot be used." % (filter_value, filter_name)
error_msg += " Use one of: [%s]" % ", ".join(map(str, dataset_filter.accepted_values))
raise biomart.BiomartException(error_msg)
# check attributes unless we're only counting
if not count:
# no attributes given, use default attributes
if not attributes:
attributes = [a.name for a in self.attributes.values() if a.is_default]
# if no default attributes have been defined, raise an exception
if not attributes:
raise biomart.BiomartException("at least one attribute is required, none given")
for attribute_name in attributes:
if attribute_name not in self.attributes.keys():
self.show_attributes()
raise biomart.BiomartException("The attribute '%s' does not exist." % attribute_name)
# selected attributes must belong to the same attribute page.
if len(set([self.attributes[a].attribute_page for a in attributes])) > 1:
self.show_attributes()
raise biomart.BiomartException("You must use attributes that belong to the same attribute page.")
# filters and attributes looks ok, start building the XML query
root = Element('Query')
root.attrib.update({
'virtualSchemaName': 'default', # TODO: use database virtualSchemaName instead (if any error)
'formatter': 'TSV',
'header': str(header),
'uniqueRows': '1',
'datasetConfigVersion': '0.6',
'count': count is True and '1' or ''
})
dataset = SubElement(root, "Dataset")
dataset.attrib.update({
'name': self.name,
'interface': self.interface
})
# Add filters to the XML query
for filter_name, filter_value in filters.items():
dataset_filter = self.filters[filter_name]
filter_elem = SubElement(dataset, "Filter")
filter_elem.set('name', filter_name)
if 'boolean_list' == dataset_filter.filter_type:
if filter_value is True or filter_value.lower() in ('included', 'only'):
filter_elem.set('excluded', '0')
elif filter_value is False or filter_value.lower() == 'excluded':
filter_elem.set('excluded', '1')
else:
if isinstance(filter_value, list) or isinstance(filter_value, tuple):
filter_value = ",".join(map(str, filter_value))
filter_elem.set('value', str(filter_value))
# Add attributes to the XML query, unless we're only counting
if not count:
for attribute_name in attributes:
attribute_elem = SubElement(dataset, "Attribute")
attribute_elem.set('name', str(attribute_name))
if self.verbose:
print("[BiomartDataset] search query:\n%s" % tostring(root))
return self.server.get_request(query = tostring(root))