def cli():
"""Create command line interface and parse arguments for main program."""
parser = argparse.ArgumentParser(
description=__doc__,
formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument(
"-i",
"--fast5_dir",
action=FullPathsList,
help="""Directory of fast5 files you want to query. Program will
walk recursively through subdirectories.""",
type=str,
nargs="+",
required=True)
parser.add_argument("-r",
"--reference",
action=FullPathsList,
help="""Fastq or BAM/SAM file(s).""",
nargs="+",
required=True)
parser.add_argument(
"-o",
"--output",
action=FullPaths,
help="""Filename to write fast5 paths to. If nothing is entered,
it will write the paths to STDOUT.""",
default=None)
parser.add_argument(
"-m",
"--mapped",
action='store_true',
help="Only extract read ids for mapped reads in BAM/SAM files.",
default=False)
parser.add_argument(
"--log_level",
help="Level of logging. 0 is none, 5 is for debugging. Default is 4 "
"which will report info, warnings, errors, and critical "
"information.",
default=4,
type=int,
choices=range(6))
parser.add_argument("--no_progress_bar",
help="Do not display progress bar.",
action='store_true',
default=False)
args = parser.parse_args()
setup_logging(args.log_level)
if args.output is not None:
args.output = open(args.output, 'w')
logging.info(" Starting fast5seek.")
fast5seek.main(args)
logging.info(" Done.")
def test_EcoliTestFast5DataSam_TwoFilepaths(self):
fast5_dir = ['tests/data/fast5']
reference = ['tests/data/sam/ecoli.sam']
args = TestArgs(fast5_dir, reference, True)
f = io.StringIO()
with redirect_stdout(f):
fast5seek.main(args)
stdout = f.getvalue()
result = [x for x in stdout.split('\n') if x]
expected = [
'tests/data/fast5/ecoli1.fast5', 'tests/data/fast5/ecoli2.fast5'
]
self.assertCountEqual(result, expected)
def test_TBTestFast5DataBam_SixFilepaths(self):
fast5_dir = ['tests/data/fast5']
reference = ['tests/data/bam/tb.bam']
args = TestArgs(fast5_dir, reference, True)
f = io.StringIO()
with redirect_stdout(f):
fast5seek.main(args)
stdout = f.getvalue()
result = [x for x in stdout.split('\n') if x]
expected = [
'tests/data/fast5/tb5.fast5', 'tests/data/fast5/tb6.fast5',
'tests/data/fast5/tb4.fast5', 'tests/data/fast5/tb3.fast5',
'tests/data/fast5/tb2.fast5', 'tests/data/fast5/tb1.fast5'
]
self.assertCountEqual(result, expected)
def test_TBTestFast5DataAlbacoreFastq_EightFilepaths(self):
fast5_dir = ['tests/data/fast5']
reference = ['tests/data/fastq/basecalled.fastq.gz']
args = TestArgs(fast5_dir, reference, False)
f = io.StringIO()
with redirect_stdout(f):
fast5seek.main(args)
stdout = f.getvalue()
result = [x for x in stdout.split('\n') if x]
expected = [
'tests/data/fast5/ecoli1.fast5', 'tests/data/fast5/ecoli2.fast5',
'tests/data/fast5/tb5.fast5', 'tests/data/fast5/tb6.fast5',
'tests/data/fast5/tb4.fast5', 'tests/data/fast5/tb3.fast5',
'tests/data/fast5/tb2.fast5', 'tests/data/fast5/tb1.fast5'
]
self.assertCountEqual(result, expected)