def _get_genome_id_to_path_map(
self, file_path_of_file_mapping_genome_id_to_paths,
list_of_drawn_genome_id):
"""
Get a dictionary mapping genome id to the path of their genome
@param file_path_of_file_mapping_genome_id_to_paths: File path to file with format 'id \t path'
@type file_path_of_file_mapping_genome_id_to_paths: str | unicode
@param list_of_drawn_genome_id: List of genome identifiers
@type list_of_drawn_genome_id: list[str|unicode]
@return: genome ids mapped to their gnome file path
@rtype: dict[str|unicode, str|unicode]
"""
genome_id_to_path_map = {}
mdt = MetadataTable(logfile=self._logfile, verbose=self._verbose)
mdt.read(file_path_of_file_mapping_genome_id_to_paths)
if mdt.get_number_of_rows() > 0:
genome_id_to_path_map = mdt.get_map(0, 1, unique_key=True)
msg = "'{}' is missing one or more genome id".format(
os.path.basename(file_path_of_file_mapping_genome_id_to_paths))
assert set(genome_id_to_path_map.keys()).issuperset(
list_of_drawn_genome_id), msg
return {
genome_id: genome_id_to_path_map[genome_id]
for genome_id in list_of_drawn_genome_id
}
def create_meta_table(self, file_path_metadata_table):
"""
Generate a input metadata file with genome ids only
@param file_path_metadata_table:
@type file_path_metadata_table: str|unicode
@rtype: None
"""
metadata_table = MetadataTable(separator=self._separator, logfile=self._logfile, verbose=self._verbose)
metadata_table.read(self._file_path_reference_genome_locations, column_names=False)
if metadata_table.get_number_of_rows() == 0:
raise ValueError("Invalid file content")
id_column = metadata_table.get_column(0)
metadata_table.clear()
metadata_table.insert_column(id_column, self._column_name_genome_id)
metadata_table.write(file_path_metadata_table, column_names=True)
def _get_genome_id_to_path_map(self, file_path):
"""
Get a map of genome_id to genome path
@param file_path: File path
@type file_path: str | unicode
@return: map of genome_id to genome path
@rtype: dict[str|unicode, str|unicode]
"""
assert self.validate_file(file_path)
data_table = MetadataTable(separator=self._separator,
logfile=self._logfile,
verbose=self._verbose)
data_table.read(file_path, column_names=False)
if data_table.get_number_of_rows() == 0:
self._logger.warning("No data in file '{}'.".format(file_path))
return {}
dict_genome_id_to_path = data_table.get_map(0, 1)
return dict_genome_id_to_path