def get_arguments():
"""Create the command line interface and return the command line arguments
Returns
-------
Namespace
The command line arguments
"""
parser = argparse.ArgumentParser(
formatter_class=argparse.RawTextHelpFormatter,
description="Example Command:\n"
"--json-file=data/run1.B1phi1.i1.seqwish.w100.json --cells-per-file=5000 --fasta=data/run1.B1phi1.i1.seqwish.fasta"
)
parser.add_argument('-j',
'--json-file',
dest='json_file',
required=True,
help='input JSON file')
parser.add_argument(
'-f',
'--fasta',
dest='fasta',
help=
'Optional: Fasta file containing the pangenome sequence generated by '
'odgi for this Graph.')
parser.add_argument('-o',
'--out-folder',
dest='output_folder',
help='output folder')
parser.add_argument(
'-c',
'--cells-per-file',
dest='cells_per_file',
default=5000,
type=int,
help='Tip: Adjust this number to get chunk files output close to 2MB. '
'Number of cells per file (#bins per file = #cells / #paths)')
parser.add_argument(
'-l',
'--log-level',
default='DEBUG',
choices=('DEBUG', 'INFO', 'WARNING', 'ERROR'),
help='level of logging verbosity. DEBUG is most verbose')
args = parser.parse_args()
if not args.output_folder:
# directory with the same name as the json
args.output_folder = osPath(args.json_file).parent.joinpath(
osPath(args.json_file).stem)
else:
args.output_folder = osPath(args.output_folder)
os.makedirs(args.output_folder, exist_ok=True)
return args
def write_json_files(json_file, schematic: PangenomeSchematic):
partitions, bin2file_mapping = schematic.split(args.cells_per_file)
folder = osPath(json_file).parent
if args.output_folder:
folder = osPath(args.output_folder)
os.makedirs(folder, exist_ok=True) # make directory for all files
for part in partitions:
p = folder.joinpath(part.filename)
with p.open('w') as fpgh9:
fpgh9.write(part.json_dump())
print("Saved results to", p)
schematic.write_index_file(folder, bin2file_mapping)
def setup_logging():
"""Setup the logging, add a log file"""
log_name = osPath(args.json_file).with_suffix('.log')
if args.output_folder:
log_name = osPath(args.output_folder).joinpath('log')
os.makedirs(args.output_folder, exist_ok=True)
t = datetime.now()
timestr = f"{t.year}{t.month:02}{t.day:02}-{t.hour:02}:{t.minute:02}:{t.second:02}"
log_name = str(log_name) + '.' + timestr
handler = logging.FileHandler(log_name)
handler.setLevel(args.log_level)
handler.setFormatter(
logging.Formatter(matrixcomponent.LOGGING_FORMAT_STR,
datefmt=matrixcomponent.LOGGING_DATE_FORMAT))
logging.getLogger().addHandler(handler)
def main():
global args
args = get_arguments()
setup_logging()
if args.parallel_cores > 0:
chunk_size = args.parallel_cores
else:
chunk_size = os.cpu_count()
parallel = Parallel(n_jobs=chunk_size, prefer="processes")
if args.json_file.endswith("*"):
files = glob(args.json_file + '.json')
print("===Input Files Found===\n", '\n'.join(files))
else:
files = [args.json_file]
for json_file in files:
LOGGER.info(f'reading {osPath(json_file)}...\n')
paths, pangenome_length, bin_width = JSONparser.parse(
json_file, chunk_size * 2, parallel) # give 2x jobs to do
schematic = segment_matrix(paths, bin_width, args.cells_per_file,
pangenome_length, args.no_adjacent_links,
parallel)
# this one spits out json and optionally other output files (fasta, ttl)
path_name = str(bin_width)
folder_path = osPath(args.output_folder).joinpath(
path_name) # full path
write_files(folder_path, args.fasta, schematic, args.no_adjacent_links)
LOGGER.info("Finished processing the file " + json_file)
def resource_directory(_context, name, directory):
if directory:
directory = osPath(_context.path(directory))
if not directory.is_dir():
raise ConfigurationError('No such directory', directory)
root = getUtility(IApplication, 'root')
if name not in root:
# prevent circular import
from plone.server.content import StaticDirectory
root[name] = StaticDirectory(directory)
def resource_directory(_context, name, directory):
if directory:
directory = osPath(_context.path(directory))
if not directory.is_dir():
raise ConfigurationError('No such directory', directory)
root = getUtility(IApplication, 'root')
if name not in root:
# prevent circular import
from plone.server.content import StaticDirectory
root[name] = StaticDirectory(directory)
def get_arguments():
"""Create the command line interface and return the command line arguments
Returns
-------
Namespace
The command line arguments
"""
parser = argparse.ArgumentParser(
formatter_class=argparse.RawTextHelpFormatter,
description="Example Command:\n"
"--json-file=data/run1.B1phi1.i1.seqwish.w100.json --cells-per-file=5000 --fasta=data/run1.B1phi1.i1.seqwish.fasta"
)
parser.add_argument('-j',
'--json-file',
dest='json_file',
required=True,
help='input JSON file')
parser.add_argument(
'-f',
'--fasta',
dest='fasta',
help=
'Optional: Fasta file containing the pangenome sequence generated by '
'odgi for this Graph.')
parser.add_argument('-o',
'--out-folder',
dest='output_folder',
help='output folder')
parser.add_argument(
'-c',
'--cells-per-file',
dest='cells_per_file',
default=5000,
type=int,
help='Tip: Adjust this number to get chunk files output close to 2MB. '
'Number of cells per file (#bins per file = #cells / #paths)')
parser.add_argument(
'-l',
'--log-level',
default='DEBUG',
choices=('DEBUG', 'INFO', 'WARNING', 'ERROR'),
help='level of logging verbosity. DEBUG is most verbose')
parser.add_argument(
'-p',
'--parallel-cores',
dest='parallel_cores',
default=os.cpu_count(),
type=int,
help='Tip: do not set this one to more than available CPU cores)')
parser.add_argument(
'-nal',
'--no-adjacent-links',
dest='no_adjacent_links',
default=False,
action='store_true',
help='Switches off the add_adjacent_connector_column() routine)')
args = parser.parse_args()
# file path logic for single or list of files with wildcard *
if not args.output_folder:
if args.json_file.endswith("*"): # directory is user provided prefix
args.output_folder = args.json_file[:-1]
elif args.json_file.endswith(".json"): # single json file
args.output_folder = osPath(args.json_file).parent.joinpath(
osPath(args.json_file).stem)
else:
print(
"Please provide an --out-folder or end --json-file= prefix with a *",
file=sys.stderr)
exit(1)
else:
args.output_folder = osPath(args.output_folder)
os.makedirs(args.output_folder, exist_ok=True)
if args.parallel_cores <= 0:
args.parallel_cores = os.cpu_count()
return args
