def __init__(self, database, overwrite=None, clean=True):
'''
@param: database
@pdef: database to blast upon.
@ptype: {String}
@param: overwrite
@pdef: For writing actions. Decides whether it can overwrite an
existing file.
@pdefault: _SBIglobals.overwrite_
@ptype: {Boolean}
@param: clean
@pdef: remove the temporary files after the data is read.
@pdefault: _True_
@pclash: if _SBIglobals.debug_ is _True_, clean is _False_
@ptype: {Boolean}
@raises: {HmmError}
'''
self._error = HmmError()
#hmmer executable configuration
if HmmExe._EXE is None:
self._set_default_executable('hmmer')
HmmExe._DBFORMATER = HmmExe._CONFIG.get('hmmer', 'dbformatexe')
# Local overwrite takes precedence over Global overwrite
self._overwrite = SBIg.decide_overwrite(overwrite)
self._database = self._check_database(os.path.abspath(database))
self._clean_files = clean
#Optional execution parameters
self._parameters = {'attr': [], 'flag': []}
def write(self, output_file=None, format='PDB', force=None, clean=False):
"""
Writes the object in a specific format
@type output_file: String
@param output_file: File to write
@type format: String
@param format: Format of the file to print
"""
outfile = File(
file_name=output_file, action='w', overwrite=SBIg.decide_overwrite(force))
if format == 'PDB':
self._write_PDB_file(pdb_file=outfile, clean=clean)
def __init__(self, file_name=None, action='r', overwrite=None):
if file_name is None:
raise FileError(0) # Raise 'No file specified' Error
self._file = file_name
SBIglobals.alert('debug', self,
'Preparing File: {0}'.format(self.full))
self._action = None
self._check_action(action.lower()) #action must be valid
self._fd = None
# Local overwrite takes precedence over Global overwrite
self._overwrite = SBIglobals.decide_overwrite(overwrite)
self._check_file()
def copy_file(source, destination, overwrite=False):
"""
- copy_file() : Copy a file from one place to another.
- source (string): Name of the original file
@Mandatory
- destination (string): Name of the new copied file
@Mandatory
-overwrite (bool) : Ignores if a file with the same name existed before
@Default: False
@Raises FileError
"""
# Local overwrite takes precedence over Global overwrite
overwrite = SBIglobals.decide_overwrite(overwrite)
if not os.path.exists(source):
raise FileError(3, source, 'noexists')
if os.path.exists(destination) and not overwrite:
raise FileError(3, destination, 'exists')
shutil.copyfile(src=source, dst=destination)
def __init__(self, database, search_type = 'protein', self_hit = False,
hitid_format = 'single', overwrite = None, clean = True):
'''
@param: database
@pdef: database to blast upon.
@ptype: {String}
@param: search_type
@pdef: type of database search.
@pdefault: 'protein'
@ptype: {String}
@@param: self_hit
@pdef: when _True_ if the query is found in the database, it is
retrieved.
@pdefault: _False_
@ptype: {Boolean}
@param: hitid_format
@pdef: format of the name of the hit. If given a wrong option,
it defaults to 'single'
@pdefault: 'single'
@poptions: 'single' -> first word of the name,
'double' -> first two words of the hit name,
'all' -> all the text in the hit name
@ptype: {String}
@param: overwrite
@pdef: For writing actions. Decides whether it can overwrite an
existing file.
@pdefault: _SBIglobals.overwrite_
@ptype: {Boolean}
@param: clean
@pdef: remove the temporary files after the data is read.
@pdefault: _True_
@pclash: if _SBIglobals.debug_ is _True_, clean is _False_
@ptype: {Boolean}
@raises: {BlastError}
'''
self._error = BlastError()
self._search_type = self._check_execution_type(search_type)
# Blast executable configuration
if self._EXE is None:
self._set_default_executable('blast')
BlastExe._DBFORMATER = BlastExe._CONFIG.get('blast', 'dbformatexe')
# Database Configuration
self._idx = None
self._database = self._check_database(os.path.abspath(database))
# Optional execution parameters
self._parameters = []
msg = ['New Blast Executable created.', ]
msg.append('Blast executable at {0}'.format(self._EXE.full_executable))
SBIg.alert('debug', self, msg)
self._self_hit = self_hit
self._hitid_format = self._check_hitid_format(hitid_format)
self._clean_files = clean
# Local overwrite takes precedence over Global overwrite
self._overwrite = SBIg.decide_overwrite(overwrite)
def split(self,
start_separator=None,
end_separator=None,
size=1073741824,
sections=None,
overwrite=None):
"""
> split() : Splits a given file in multiple files
- start_separator (string): Line startswith. Is a line to mark the line that starts the new file
@Default: None, only other parameters are considered
- end_separator (string): Line startswith. Is a line to mark the line that ends the new file
@Default: None, only other parameters are considered
- size (int): Max allowed file size. IN BYTES
@Default: 1073741824 (1GB)
- sections (int): Number of sections to split file into. Overwrites 'size'
@Default: None
- overwrite (bool): Overwrite previous files of the same name
@Returns an array with the File object of each created file.
@Raises FileError
@Raises AttributeError if both start_separator and end_separator are defined
"""
# A defined separator will avoid splitting data
# For example, uniprot files separate between proteins with '//', then by declaring the separator
# we can avoid a protein from being split between two files.
counter = 1
newfiles = []
if (size is None and sections is None) or not isinstance(size, int):
raise FileError(5, self.__name__, 'split')
if (start_separator is not None and end_separator is not None):
raise AttributeError(
'Both start and end separators can not be defined simultaniously\n'
)
# section TAKES PRECEDENCE over size...
if sections is not None:
size = math.ceil(os.path.getsize(self.full) / float(sections))
else:
sections = math.ceil(os.path.getsize(self.full) / size)
# preparing partition file names
outputfile = os.path.join(self.dir,
self.prefix + '.{0:003d}' + self.extension)
if self.is_gziped: outputfile += '.gz'
# Local overwrite takes precedence over Global overwrite
overwrite = SBIglobals.decide_overwrite(overwrite)
SBIglobals.alert(
'verbose', self,
'Dividing {0.full} into {1} files (aprox.)'.format(self, sections))
newfiles.append(
File(file_name=outputfile.format(counter),
action='w',
overwrite=overwrite))
for line in self.descriptor:
if (start_separator is None or line.startswith(start_separator)
) and os.path.getsize(newfiles[-1].full) >= size:
newfiles[-1].close()
counter += 1
newfiles.append(
File(file_name=outputfile.format(counter),
action='w',
overwrite=overwrite))
newfiles[-1].write(line)
if (end_separator is None or line.startswith(end_separator)
) and os.path.getsize(newfiles[-1].full) >= size:
newfiles[-1].close()
counter += 1
newfiles.append(
File(file_name=outputfile.format(counter),
action='w',
overwrite=overwrite))
newfiles[-1].close()
self.close()
return newfiles
