def main(args):
if args.split:
kevlar.mkdirp(args.split, trim=True)
outstream = None if args.split else kevlar.open(args.out, 'w')
readstream = kevlar.parse_augmented_fastx(kevlar.open(args.infile, 'r'))
partitioner = partition(readstream,
strict=args.strict,
minabund=args.min_abund,
maxabund=args.max_abund,
dedup=args.dedup,
gmlfile=args.gml,
logstream=args.logfile)
partnum = 0
numreads = 0
for partnum, part in enumerate(partitioner, 1):
numreads += len(part)
if args.split:
ofname = '{:s}.cc{:d}.augfastq.gz'.format(args.split, partnum)
with kevlar.open(ofname, 'w') as outfile:
for read in part:
kevlar.print_augmented_fastx(read, outfile)
else:
for read in part:
read.name += ' kvcc={:d}'.format(partnum)
kevlar.print_augmented_fastx(read, outstream)
message = '[kevlar::partition] grouped {:d} reads'.format(numreads)
message += ' into {:d} connected components'.format(partnum)
print(message, file=args.logfile)
def test_mkdirp():
tempdir = tempfile.mkdtemp()
path1 = os.path.join(tempdir, 'partitions')
assert kevlar.mkdirp(path1) == path1
path2 = os.path.join(tempdir, 'partitions2', 'part')
path2test = os.path.join(tempdir, 'partitions2')
assert kevlar.mkdirp(path2, trim=True) == path2test
path3 = os.path.join(tempdir, 'partitions3', 'a', 'long', 'path', 'PART')
path3test = os.path.join(tempdir, 'partitions3', 'a', 'long', 'path')
assert kevlar.mkdirp(path3, trim=True) == path3test
shutil.rmtree(tempdir)