def parseHiC(filename, **kwargs):
"""Returns an :class:`.HiC` from a Hi-C data file.
This function extends :func:`.parseHiCStream`.
:arg filename: the filename to the Hi-C data file.
:type filename: str
"""
import os, struct
title = kwargs.get('title')
if title is None:
title = os.path.basename(filename)
else:
title = kwargs.pop('title')
if isURL(filename):
hic = parseHiCBinary(filename, title=title, **kwargs)
else:
with open(filename,'rb') as req:
magic_number = struct.unpack('<3s',req.read(3))[0]
if magic_number == b"HIC":
hic = parseHiCBinary(filename, title=title, **kwargs)
else:
with open(filename, 'r') as filestream:
hic = parseHiCStream(filestream, title=title, **kwargs)
return hic
def update(self, source=None):
"""Update data and files from CATH."""
self._source = source = self._source or source
self.reset()
if source is None:
return
LOGGER.timeit('_cath_update')
type_ = 0
tree = None
if isinstance(source, str):
if isfile(source):
type_ = 1
elif isURL(source):
type_ = 0
else:
type_ = 2
elif hasattr(source, 'read'):
type_ = 1
else:
raise TypeError(
'source must be either an url, file name, file handle, '
'or text in xml format')
if type_ == 0:
LOGGER.info('Fetching data from CATH...')
self._fetch()
LOGGER.info('Parsing CATH files...')
self._parse()
elif type_ == 1:
LOGGER.info('Reading data from the local xml file...')
tree = ET.parse(source)
elif type_ == 2:
LOGGER.info('Parsing input string...')
tree = ET.fromstring(source)
# post-processing
if type_ > 0:
root = tree.getroot()
nodes = root.iter()
# remove prefix from node tags
for node in nodes:
node.tag = node.tag.lstrip('id.')
# convert int to str
length_nodes = root.findall('.//*[@length]')
for node in length_nodes:
node.attrib['length'] = int(node.attrib['length'])
copy2(root, self.root)
self._update_map()
LOGGER.report('CATH local database built in %.2fs.', '_cath_update')
def update(self, source=None):
"""Update data and files from CATH."""
self._source = source = self._source or source
self.reset()
if source is None:
return
LOGGER.timeit('_cath_update')
type_ = 0
tree = None
if isinstance(source, str):
if isfile(source):
type_ = 1
elif isURL(source):
type_ = 0
else:
type_ = 2
elif hasattr(source, 'read'):
type_ = 1
else:
raise TypeError('source must be either an url, file name, file handle, '
'or text in xml format')
if type_ == 0:
LOGGER.info('Fetching data from CATH...')
self._fetch()
LOGGER.info('Parsing CATH files...')
self._parse()
elif type_ == 1:
LOGGER.info('Reading data from the local xml file...')
tree = ET.parse(source)
elif type_ == 2:
LOGGER.info('Parsing input string...')
tree = ET.fromstring(source)
# post-processing
if type_ > 0:
root = tree.getroot()
nodes = root.iter()
# remove prefix from node tags
for node in nodes:
node.tag = node.tag.lstrip('id.')
# convert int to str
length_nodes = root.findall('.//*[@length]')
for node in length_nodes:
node.attrib['length'] = int(node.attrib['length'])
copy2(root, self.root)
self._update_map()
LOGGER.report('CATH local database built in %.2fs.', '_cath_update')