def get_parser():
"""
returns the parser object for the oxli subcommand handler
"""
parser = argparse.ArgumentParser(
description='Single entry point script for khmer',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
subparsers = parser.add_subparsers()
# build-graph (formerly load-graph.py) parsers here
parser_build_graph = \
subparsers.add_parser('build-graph',
help="Load sequences into the compressible graph"
"format plus optional tagset",
description="Load sequences into the "
"compressible graph format plus optional tagset")
khmer_args.build_nodegraph_args("Load sequences into the compressible"
"graph format plus optional tagset.",
None, parser=parser_build_graph)
build_graph.build_parser(parser_build_graph)
parser_build_graph.set_defaults(func=build_graph.main)
return parser
def get_parser():
parser = build_nodegraph_args('Takes a partitioned reference file \
and a list of reads, and sorts reads \
by which partition they connect to')
parser.epilog = EPILOG
parser.add_argument('-r',
'--traversal_range',
type=int,
dest='traversal_range',
default=DEFAULT_RANGE,
help='depth of breadth-first search to perform\
from each read')
parser.add_argument('--max_queue_size', type=int, default=1000)
parser.add_argument('--prefix',
dest='output_prefix',
default=DEFAULT_OUT_PREF,
help='Prefix for sorted read files')
parser.add_argument('--outdir',
dest='outdir',
default='',
help='output directory; default is location of \
fastp file')
parser.add_argument('--query',
dest='query',
nargs='+',
help='Reads to be swept and sorted')
parser.add_argument('--db',
dest='db',
nargs='+',
help='Database reads for sweep',
required=True)
return parser
def get_parser():
epilog = """\
Load in a set of sequences, partition them, merge the partitions, and
annotate the original sequences files with the partition information.
This script combines the functionality of
:program:`load-graph.py`, :program:`partition-graph.py`,
:program:`merge-partitions.py`, and :program:`annotate-partitions.py` into
one script. This is convenient but should probably not be used for large
data sets, because :program:`do-partition.py` doesn't provide save/resume
functionality.
Example::
do-partition.py -k 20 example tests/test-data/random-20-a.fa
"""
parser = build_nodegraph_args(
descr='Load, partition, and annotate FAST[AQ] sequences',
epilog=textwrap.dedent(epilog))
add_threading_args(parser)
parser.add_argument('--subset-size', '-s', default=DEFAULT_SUBSET_SIZE,
dest='subset_size', type=float,
help='Set subset size (usually 1e5-1e6 is good)')
parser.add_argument('--no-big-traverse', dest='no_big_traverse',
action='store_true', default=False,
help='Truncate graph joins at big traversals')
parser.add_argument('--keep-subsets', dest='remove_subsets',
default=True, action='store_false',
help='Keep individual subsets (default: False)')
parser.add_argument('graphbase', help="base name for output files")
parser.add_argument('input_filenames', metavar='input_sequence_filename',
nargs='+', help='input FAST[AQ] sequence filenames')
parser.add_argument('-f', '--force', default=False, action='store_true',
help='Overwrite output file if it exists')
return parser
def get_parser():
parser = build_nodegraph_args(descr="Load sequences into the compressible "
"graph format plus optional tagset.",
citations=['graph', 'SeqAn'])
parser = build_graph.build_parser(parser)
return parser
def main():
parser = khmer_args.build_nodegraph_args()
parser.add_argument("--samples", nargs="+")
parser.add_argument("--save-prefix")
parser.add_argument("--print-tree", action="store_true", default=False)
args = parser.parse_args()
if not args.save_prefix:
print >> sys.stderr, "No save prefix specified! Exiting..."
sys.exit(1)
factory = NodegraphFactory(args)
root = sbt.Node(factory)
for sample_fn in args.samples:
print "*** Build node for", sample_fn
leaf = sbt.Leaf(os.path.basename(sample_fn), os.path.basename(sample_fn), factory.create_nodegraph())
print "--- Consuming file..."
leaf.graph.consume_fasta(sample_fn)
print "--- Adding node to SBT..."
root.add_node(leaf)
print "--- Done with", sample_fn
if args.print_tree:
sbt.print_sbt(root)
print "\n*** Saving to disk"
fn = sbt.save_sbt(root, args.save_prefix)
print "--- Save to", fn
def get_parser():
parser = build_nodegraph_args(descr="Load sequences into the compressible "
"graph format plus optional tagset.")
add_threading_args(parser)
parser.add_argument('input_filenames', metavar='input_sequence_filename',
nargs='+', help='input FAST[AQ] sequence filename')
return parser
def get_parser():
parser = build_nodegraph_args(descr="Load sequences into the compressible "
"graph format plus optional tagset.")
add_threading_args(parser)
parser.add_argument('input_filenames', metavar='input_sequence_filename',
nargs='+', help='input FAST[AQ] sequence filename')
return parser
def main():
parser = build_nodegraph_args()
parser.add_argument('-o',
'--outfile',
help='output file; default is "infile".sweep2')
parser.add_argument('-q', '--quiet')
parser.add_argument('input_filename')
parser.add_argument('read_filename')
args = parser.parse_args()
inp = args.input_filename
readsfile = args.read_filename
outfile = os.path.basename(readsfile) + '.sweep2'
if args.outfile:
outfile = args.outfile
outfp = open(outfile, 'w')
# create a nodegraph data structure
ht = khmer_args.create_countgraph(args)
# load contigs, connect into N partitions
print('loading input reads from', inp)
ht.consume_seqfile(inp)
print('starting sweep.')
m = 0
K = ht.ksize()
instream = screed.open(readsfile)
for n, is_pair, read1, read2 in broken_paired_reader(instream):
if n % 10000 == 0:
print('...', n, m)
if is_pair:
count1 = ht.get_median_count(read1.sequence)[0]
count2 = ht.get_median_count(read2.sequence)[0]
if count1 or count2:
m += 1
write_record_pair(read1, read2, outfp)
else:
count = ht.get_median_count(read1.sequence)[0]
if count:
m += 1
write_record(read1, outfp)
def main():
parser = build_nodegraph_args()
parser.add_argument('-o', '--outfile',
help='output file; default is "infile".sweep2')
parser.add_argument('-q', '--quiet')
parser.add_argument('input_filename')
parser.add_argument('read_filename')
args = parser.parse_args()
inp = args.input_filename
readsfile = args.read_filename
outfile = os.path.basename(readsfile) + '.sweep2'
if args.outfile:
outfile = args.outfile
outfp = open(outfile, 'w')
# create a nodegraph data structure
ht = khmer_args.create_countgraph(args)
# load contigs, connect into N partitions
print('loading input reads from', inp)
ht.consume_fasta(inp)
print('starting sweep.')
m = 0
K = ht.ksize()
instream = screed.open(readsfile)
for n, is_pair, read1, read2 in broken_paired_reader(instream):
if n % 10000 == 0:
print('...', n, m)
if is_pair:
count1 = ht.get_median_count(read1.sequence)[0]
count2 = ht.get_median_count(read2.sequence)[0]
if count1 or count2:
m += 1
write_record_pair(read1, read2, outfp)
else:
count = ht.get_median_count(read1.sequence)[0]
if count:
m += 1
write_record(read1, outfp)
def get_parser():
"""Return the parser object for the oxli subcommand handler."""
parser = argparse.ArgumentParser(
description='Single entry point script for khmer',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
subparsers = parser.add_subparsers()
# build-graph (formerly load-graph.py) parsers here
parser_build_graph = \
subparsers.add_parser(
name='build-graph',
help="Load sequences into the compressible graph format "
"plus optional tagset")
parser_build_graph = build_nodegraph_args(parser=parser_build_graph)
build_graph.build_parser(parser_build_graph)
parser_build_graph.set_defaults(func=build_graph.main)
return parser
def get_parser():
parser = build_nodegraph_args('Takes a partitioned reference file \
and a list of reads, and sorts reads \
by which partition they connect to')
parser.epilog = EPILOG
parser.add_argument(
'-r', '--traversal_range', type=int, dest='traversal_range',
default=DEFAULT_RANGE, help='depth of breadth-first search to perform\
from each read')
parser.add_argument('-b', '--buffer_size', dest='max_reads', type=int,
default=DEFAULT_MAX_READS,
help='Max total reads to buffer before flushing')
parser.add_argument('-l', '--buffer_length', dest='buffer_size', type=int,
default=DEFAULT_BUFFER_SIZE,
help='Max length of an individual label buffer \
before flushing')
parser.add_argument('--prefix', dest='output_prefix',
default=DEFAULT_OUT_PREF,
help='Prefix for sorted read files')
parser.add_argument('--outdir', dest='outdir',
help='output directory; default is location of \
fastp file')
parser.add_argument('-m', '--max_buffers', dest='max_buffers', type=int,
default=DEFAULT_NUM_BUFFERS,
help='Max individual label buffers before flushing')
labeling = parser.add_mutually_exclusive_group(required=True)
labeling.add_argument('--label-by-pid', dest='label_by_pid',
action='store_true', help='separate reads by\
reference partition id')
labeling.add_argument('--label-by-seq', dest='label_by_seq',
action='store_true', help='separate reads by\
reference sequence')
labeling.add_argument('--label-by-group', dest='group_size', type=int,
help='separate reads by arbitrary sized groups\
of reference sequences')
parser.add_argument(dest='input_fastp', help='Reference fasta or fastp')
parser.add_argument('input_files', nargs='+',
help='Reads to be swept and sorted')
parser.add_argument('-f', '--force', default=False, action='store_true',
help='Overwrite output file if it exists')
return parser
def get_parser():
parser = build_nodegraph_args('Takes a partitioned reference file \
and a list of reads, and sorts reads \
by which partition they connect to')
parser.epilog = EPILOG
parser.add_argument(
'-r', '--traversal_range', type=int, dest='traversal_range',
default=DEFAULT_RANGE, help='depth of breadth-first search to perform\
from each read')
parser.add_argument('-b', '--buffer_size', dest='max_reads', type=int,
default=DEFAULT_MAX_READS,
help='Max total reads to buffer before flushing')
parser.add_argument('-l', '--buffer_length', dest='buffer_size', type=int,
default=DEFAULT_BUFFER_SIZE,
help='Max length of an individual label buffer \
before flushing')
parser.add_argument('--prefix', dest='output_prefix',
default=DEFAULT_OUT_PREF,
help='Prefix for sorted read files')
parser.add_argument('--outdir', dest='outdir',
help='output directory; default is location of \
fastp file')
parser.add_argument('-m', '--max_buffers', dest='max_buffers', type=int,
default=DEFAULT_NUM_BUFFERS,
help='Max individual label buffers before flushing')
labeling = parser.add_mutually_exclusive_group(required=True)
labeling.add_argument('--label-by-pid', dest='label_by_pid',
action='store_true', help='separate reads by\
reference partition id')
labeling.add_argument('--label-by-seq', dest='label_by_seq',
action='store_true', help='separate reads by\
reference sequence')
labeling.add_argument('--label-by-group', dest='group_size', type=int,
help='separate reads by arbitrary sized groups\
of reference sequences')
parser.add_argument(dest='input_fastp', help='Reference fasta or fastp')
parser.add_argument('input_files', nargs='+',
help='Reads to be swept and sorted')
parser.add_argument('-f', '--force', default=False, action='store_true',
help='Overwrite output file if it exists')
return parser
def get_parser():
parser = build_nodegraph_args('Takes a partitioned reference file \
and a list of reads, and sorts reads \
by which partition they connect to')
parser.epilog = EPILOG
parser.add_argument(
'-r', '--traversal_range', type=int, dest='traversal_range',
default=DEFAULT_RANGE, help='depth of breadth-first search to perform\
from each read')
parser.add_argument('--max_queue_size', type=int, default=1000)
parser.add_argument('--prefix', dest='output_prefix',
default=DEFAULT_OUT_PREF,
help='Prefix for sorted read files')
parser.add_argument('--outdir', dest='outdir', default='',
help='output directory; default is location of \
fastp file')
parser.add_argument('--query', dest='query', nargs='+',
help='Reads to be swept and sorted')
parser.add_argument('--db', dest='db', nargs='+',
help='Database reads for sweep', required=True)
return parser
def main():
parser = khmer_args.build_nodegraph_args()
parser.add_argument('--samples', nargs='+')
parser.add_argument('--save-prefix')
parser.add_argument('--print-tree', action='store_true', default=False)
args = parser.parse_args()
if not args.save_prefix:
print('No save prefix specified! Exiting...', file=sys.stderr)
sys.exit(1)
factory = NodegraphFactory(args)
tree = sbt.SBT(factory)
for sample_fn in args.samples:
print('*** Build node for', sample_fn)
leaf = sbt.Leaf(os.path.basename(sample_fn),
factory.create_nodegraph())
fname = os.path.join('.sbt.' + args.save_prefix,
".".join([args.save_prefix,
os.path.basename(sample_fn),
'sbt']))
if os.path.exists(fname):
print('--- Loading existing file...')
leaf.graph.load(fname)
else:
print('--- Consuming file...')
leaf.graph.consume_fasta(sample_fn)
print('--- Adding node to SBT...')
tree.add_node(leaf)
print('--- Done with', sample_fn)
if args.print_tree:
tree.print()
print('\n*** Saving to disk')
fn = tree.save(args.save_prefix)
print('--- Save to', fn)
def main():
parser = build_nodegraph_args()
parser.add_argument('readfile1', help='fasta sequence file to be loaded in hashtable, use "-" if from stdin')
parser.add_argument('readfile2', help='fasta readfile to query against hashtable, use "-" if from stdin')
parser.add_argument('--shared', help='shared kmer in readfile 1 and 2')
parser.add_argument('--uniq2', help='uniq kmer in readfile2')
parser.add_argument('--x2', default='1e8', help='max_table size for readfile2')
parser.add_argument('--N2', default='4', help='# of table (N) for readfile2')
args = parser.parse_args()
print(args)
K = args.ksize
HT_SIZE = args.max_tablesize
N_HT = args.n_tables
HT_SIZE2 = int(float(args.x2))
N_HT2 = int(args.N2)
# positional
readfile1 = args.readfile1
readfile2 = args.readfile2
shared = args.shared
uniq2 = args.uniq2
if readfile1 == '-' and readfile2 == '-':
mes = ('*** Only one of readfile1 and readfile2 '
'can be read from stdin')
print(mes, file=sys.stderr)
try:
if readfile1 == '-':
fp1 = sys.stdin
else:
fp1 = open(readfile1)
if readfile2 == '-':
fp2 = sys.stdin
else:
fp2 = open(readfile2)
if uniq2:
fw2 = open(uniq2, 'w')
# create a hashbits data structure
ht = khmer.Nodetable(K, HT_SIZE, N_HT)
# load contigs, connect into N partitions
print('loading input reads from {}..'.format(os.path.basename(readfile1)),
file=sys.stderr)
#ht.consume_seqfile(readfile1)
for record in fasta_iter(fp1):
ht.consume(record.sequence)
# Change 0.2 only if you really grok it. HINT: You don't.
fp_rate = khmer.calc_expected_collisions(ht)
print('fp rate estimated to be {:1.3f}'.format(fp_rate), file=sys.stderr)
if fp_rate > 0.01:
mes = ('**\n'
'** ERROR: the counting hash is too small for\n'
'** {}. Increase hashsize/num ht.\n'
'**\n'
'** Do not use these results!!')
print(mes.format(os.path.basename(readfile1)), file=sys.stderr)
sys.exit(-1)
n_unique1 = ht.n_unique_kmers()
# create a hashbits data structure
ht2 = khmer.Nodetable(K, HT_SIZE2, N_HT2)
n_unique2 = 0
n_shared = 0
for n, record in enumerate(fasta_iter(fp2)):
name = record['name']
sequence = record['sequence']
seq_len = len(sequence)
for i in range(0, seq_len + 1 - K):
kmer = sequence[i:i + K]
if (not ht2.get(kmer)):
n_unique2 += 1
if ht.get(kmer):
n_shared += 1
else:
mes = '>{}__{} length_{};k_{}\n{}\n'
fw2.write(mes.format(name, i, seq_len, K, kmer))
ht2.count(kmer)
mes = ('Unique kmer in {}:\t{}\n'
'Shared kmer:\t{}\n'
'Unique kmer in {}:\t{}\n')
print(mes.format(os.path.basename(readfile1), n_unique1,
n_shared,
os.path.basename(readfile2), n_unique2))
finally:
fp1.close()
fp2.close()
fw2.close()
def main():
parser = build_nodegraph_args("find uniq kmer in query compard to refs")
parser.add_argument(
'ref',
nargs='+',
help='fasta sequence file to be loaded in bloom filter')
parser.add_argument('--bfout',
default='nodetable.bf',
help='output bloom filter of ref')
args = parser.parse_args()
K = args.ksize
HT_SIZE = args.max_tablesize
N_HT = args.n_tables
# positional
refs = args.ref
start_time = time.time()
print('{} refs to be loaded'.format(len(refs)), file=sys.stderr)
ht = khmer.Nodetable(K, HT_SIZE, N_HT)
end_time = time.time()
secs = end_time - start_time
mes = 'initiation of bloom filter took {:.2f} hours..'
print(mes.format(secs / 3600.0), file=sys.stderr)
for index, filename in enumerate(refs):
if index != 0 and index % 100 == 0:
end_time = time.time()
secs = end_time - start_time
mes = '{} refs have been loaded with in {:.2f} hours ..'
print(mes.format(index, secs / 3600.0), file=sys.stderr)
try:
ht.consume_seqfile(filename)
except OSError as e:
mes = ('*** Skipping due to OSError (machine or system problem):'
' {}\n'
'*** Detailed error message:\n'
'*** {}')
print(mes.format(os.path.basename(filename), str(e)),
file=sys.stderr)
continue
print('Saving bloom filter to {}..'.format(args.bfout), file=sys.stderr)
ht.save(args.bfout)
# Change 0.2 only if you really grok it. HINT: You don't.
fp_rate = khmer.calc_expected_collisions(ht)
mes = 'fp rate estimated to be {:1.3f}'
print(mes.format(fp_rate), file=sys.stderr)
if fp_rate > 0.01:
mes = ('**\n'
'** ERROR: the counting hash is too small for\n'
'** refs. Increase hashsize/num ht.\n'
'**\n'
'** Do not use these results!!')
sys.exit(-1)
n_unique1 = ht.n_unique_kmers()
mes = ('Unique kmer:\t{}\n')
print(mes.format(n_unique1), file=sys.stderr)
def main():
parser = build_nodegraph_args("find uniq kmer in query compard to refs")
parser.add_argument('query',
help=('fasta readfile to query against'
'hashtable, use "-" if from stdin'))
parser.add_argument('--x2',
default='1e8',
help='max_table size for readfile2')
parser.add_argument('--N2',
default='4',
help='# of table (N) for readfile2')
parser.add_argument('--bfout', help='output bloom filter of ref')
group = parser.add_mutually_exclusive_group()
group.add_argument('--shared',
dest='output',
action='store_const',
const='shared',
help='output shared kmers')
group.add_argument('--uniq',
dest='output',
action='store_const',
const='uniq',
help='output uniq kmers in query')
group2 = parser.add_mutually_exclusive_group(required=True)
group2.add_argument(
'--ref',
nargs='+',
help='fasta sequence file to be loaded in bloom filter')
group2.add_argument('--load', help='load existing bloom filter')
parser.set_defaults(output='uniq')
args = parser.parse_args()
#print(args, file=sys.stderr)
K = args.ksize
HT_SIZE = args.max_tablesize
N_HT = args.n_tables
HT_SIZE2 = int(float(args.x2))
N_HT2 = int(args.N2)
# positional
query = args.query
output = args.output
start_time = time.time()
# load from existing bloom filter
if args.load:
print('loading bloom filter from {}..'.format(args.load),
file=sys.stderr)
ht = khmer.load_nodetable(args.load)
k = ht.ksize()
mes = ('*** incompatible ksize ({}) in {} with parameters K on '
'command line ({})')
assert k == K, mes.format(k, args.load, K)
end_time = time.time()
secs = end_time - start_time
mes = 'load bloom filter ({}) took {:.2f} hours..'
print(mes.format(os.path.basename(args.load), secs / 3600.0),
file=sys.stderr)
# create a hashbits data structure
else:
refs = args.ref
print('{} refs to be loaded'.format(len(refs)), file=sys.stderr)
if query == '-' and refs == ['-']:
print('*** query and ref can not both be "-" (read from stdin)',
file=sys.stderr)
ht = khmer.Nodetable(K, HT_SIZE, N_HT)
end_time = time.time()
secs = end_time - start_time
mes = 'initiation of bloom filter took {:.2f} hours..'
print(mes.format(secs / 3600.0), file=sys.stderr)
for index, filename in enumerate(refs):
if index != 0 and index % 100 == 0:
end_time = time.time()
secs = end_time - start_time
mes = '{} refs have been loaded with in {:.2f} hours ..'
print(mes.format(index, secs / 3600.0), file=sys.stderr)
try:
ht.consume_seqfile(filename)
except OSError as e:
mes = (
'*** Skipping due to OSError (machine or system problem):'
' {}\n'
'*** Detailed error message:\n'
'*** {}')
print(mes.format(os.path.basename(filename), str(e)),
file=sys.stderr)
continue
if args.bfout:
if args.load:
mes = '*** Bloom filter exists as {}, NOT saving again as {}..'
print(mes.format(args.load, args.bfout), file=sys.stderr)
else:
print('*** Saving bloom filter to {}..'.format(args.bfout),
file=sys.stderr)
ht.save(args.bfout)
# Change 0.2 only if you really grok it. HINT: You don't.
fp_rate = khmer.calc_expected_collisions(ht)
mes = 'fp rate estimated to be {:1.3f}'
print(mes.format(fp_rate), file=sys.stderr)
if fp_rate > 0.01:
mes = ('**\n'
'** ERROR: the counting hash is too small for\n'
'** refs. Increase hashsize/num ht.\n'
'**\n'
'** Do not use these results!!')
sys.exit(-1)
n_unique1 = ht.n_unique_kmers()
# create a hashbits data structure
ht2 = khmer.Nodetable(K, HT_SIZE2, N_HT2)
n_unique2 = 0
n_shared = 0
if output == 'uniq':
for n, record in enumerate(khmer.ReadParser(query)):
#for n, record in enumerate(screed.open(query)):
_l = record.name.split(None, 1)
if len(_l) == 2:
name, desc = _l
else:
name = _l[0]
desc = ''
sequence = record.sequence.replace('N', 'A')
seq_len = len(sequence)
if seq_len < K:
continue
for i in range(0, seq_len + 1 - K):
kmer = sequence[i:i + K]
if (not ht2.get(kmer)):
n_unique2 += 1
if ht.get(kmer):
n_shared += 1
else:
mes = '>{}__{} {}||length_{};k_{}\n{}'
print(mes.format(name, i, desc, seq_len, K, kmer))
ht2.count(kmer)
elif output == 'shared':
for n, record in enumerate(khmer.ReadParser(query)):
#for n, record in enumerate(screed.open(query)):
_l = record.name.split(None, 1)
if len(_l) == 2:
name, desc = _l
else:
name = _l[0]
desc = ''
sequence = record.sequence.replace('N', 'A')
seq_len = len(sequence)
if seq_len < K:
continue
for i in range(0, seq_len + 1 - K):
kmer = sequence[i:i + K]
if (not ht2.get(kmer)):
n_unique2 += 1
if ht.get(kmer):
n_shared += 1
mes = '>{}__{} {}||length_{};k_{}\n{}'
print(mes.format(name, i, desc, seq_len, K, kmer))
else:
pass
ht2.count(kmer)
mes = ('Unique kmer in {} (query):\t{}\n'
'Shared kmer:\t{}\n'
'Unique kmer in {}:\t{}\n')
print(mes.format(os.path.basename(query), n_unique2, n_shared, 'refs',
n_unique1),
file=sys.stderr)
def get_parser():
parser = build_nodegraph_args(
"Takes a partitioned reference file \
and a list of reads, and sorts reads \
by which partition they connect to"
)
parser.epilog = EPILOG
parser.add_argument(
"-r",
"--traversal_range",
type=int,
dest="traversal_range",
default=DEFAULT_RANGE,
help="depth of breadth-first search to perform\
from each read",
)
parser.add_argument(
"-b",
"--buffer_size",
dest="max_reads",
type=int,
default=DEFAULT_MAX_READS,
help="Max total reads to buffer before flushing",
)
parser.add_argument(
"-l",
"--buffer_length",
dest="buffer_size",
type=int,
default=DEFAULT_BUFFER_SIZE,
help="Max length of an individual label buffer \
before flushing",
)
parser.add_argument("--prefix", dest="output_prefix", default=DEFAULT_OUT_PREF, help="Prefix for sorted read files")
parser.add_argument(
"--outdir",
dest="outdir",
help="output directory; default is location of \
fastp file",
)
parser.add_argument(
"-m",
"--max_buffers",
dest="max_buffers",
type=int,
default=DEFAULT_NUM_BUFFERS,
help="Max individual label buffers before flushing",
)
labeling = parser.add_mutually_exclusive_group(required=True)
labeling.add_argument(
"--label-by-pid",
dest="label_by_pid",
action="store_true",
help="separate reads by\
reference partition id",
)
labeling.add_argument(
"--label-by-seq",
dest="label_by_seq",
action="store_true",
help="separate reads by\
reference sequence",
)
labeling.add_argument(
"--label-by-group",
dest="group_size",
type=int,
help="separate reads by arbitrary sized groups\
of reference sequences",
)
parser.add_argument(dest="input_fastp", help="Reference fasta or fastp")
parser.add_argument("input_files", nargs="+", help="Reads to be swept and sorted")
parser.add_argument("-f", "--force", default=False, action="store_true", help="Overwrite output file if it exists")
return parser
def main():
parser = build_nodegraph_args("find uniq kmer in query compard to refs")
parser.add_argument('query',
help=('fasta readfile to query against'
'hashtable, use "-" if from stdin'))
parser.add_argument('ref',
nargs='+',
help='fasta sequence file to be loaded in hashtable')
parser.add_argument('--x2',
default='1e8',
help='max_table size for readfile2')
parser.add_argument('--N2',
default='4',
help='# of table (N) for readfile2')
args = parser.parse_args()
#print(args, file=sys.stderr)
K = args.ksize
HT_SIZE = args.max_tablesize
N_HT = args.n_tables
HT_SIZE2 = int(float(args.x2))
N_HT2 = int(args.N2)
# positional
query = args.query
refs = args.ref
print('{} refs to be loaded'.format(len(refs)), file=sys.stderr)
if query == '-' and refs == ['-']:
print('*** query and ref can not both be "-" (read from stdin)',
file=sys.stderr)
# create a hashbits data structure
start_time = time.time()
ht = khmer.Nodetable(K, HT_SIZE, N_HT)
end_time = time.time()
secs = end_time - start_time
mes = 'initiation of bloom filter took {:.2f} hours..'
print(mes.format(secs / 3600.0), file=sys.stderr)
for index, filename in enumerate(refs):
if index != 0 and index % 100 == 0:
end_time = time.time()
secs = end_time - start_time
mes = '{} refs have been loaded with in {:.2f} hours ..'
print(mes.format(index, secs / 3600.0), file=sys.stderr)
try:
ht.consume_seqfile(filename)
except OSError as e:
mes = ('*** Skipping due to OSError (machine or system problem):'
' {}\n'
'*** Detailed error message:\n'
'*** {}')
print(mes.format(os.path.basename(filename), str(e)),
file=sys.stderr)
continue
# Change 0.2 only if you really grok it. HINT: You don't.
fp_rate = khmer.calc_expected_collisions(ht)
mes = 'fp rate estimated to be {:1.3f}'
print(mes.format(fp_rate), file=sys.stderr)
if fp_rate > 0.01:
mes = ('**\n'
'** ERROR: the counting hash is too small for\n'
'** refs. Increase hashsize/num ht.\n'
'**\n'
'** Do not use these results!!')
sys.exit(-1)
n_unique1 = ht.n_unique_kmers()
pair = 0
forward = 0
reverse = 0
other = 0
total_pair = 0
for n, is_pair, r1, r2 in broken_paired_reader(
khmer.ReadParser(query, require_paired=True)):
#for n, record in enumerate(screed.open(query)):
total_pair += 1
share_list = []
for record in [r1, r2]:
name, desc = record.name.split(None, 1)
sequence = record.sequence.replace('N', 'A')
seq_len = len(sequence)
if seq_len < K:
print('*** {} is shorter than {}..'.format(r1.name, K),
file=sys.stderr)
continue
for i in range(0, seq_len + 1 - K):
kmer = sequence[i:i + K]
if ht.get(kmer):
share_list.append(1)
break
else:
share_list.append(0)
if share_list == [1, 1]:
pair += 1
elif share_list == [1, 0]:
forward += 1
elif share_list == [0, 1]:
reverse += 1
else: #[0, 0]
other += 1
# do not print
continue
mes = ('>{} {}||uniq_{}\n{}\n' '>{} {}||uniq_{}\n{}')
l1 = r1.name.split(None, 1)
l2 = r2.name.split(None, 1)
print(
mes.format(l1[0], l1[1], share_list[0], r1.sequence, l2[0], l2[1],
share_list[1], r2.sequence))
mes = ('Unique kmer in ref:\t{}\n'
'Total pair:\t{}\n'
'Both primers uniq:\t{}\n'
'Pair with forward uniq:\t{}\n'
'Pair with reverse uniq:\t{}')
print(mes.format(n_unique1, total_pair, pair, forward, reverse),
file=sys.stderr)
