def main():
args = sanitize_help(get_parser()).parse_args()
htfile = args.countgraph
input_filename = args.input
output = args.output
infiles = [htfile, input_filename]
for infile in infiles:
check_input_files(infile, args.force)
check_space(infiles, args.force)
print('loading k-mer countgraph from', htfile, file=sys.stderr)
countgraph = Countgraph.load(htfile)
ksize = countgraph.ksize()
print('writing to', output.name, file=sys.stderr)
output = csv.writer(output)
# write headers:
output.writerow(['name', 'median', 'average', 'stddev', 'seqlen'])
for record in screed.open(input_filename):
seq = record.sequence.upper()
if 'N' in seq:
seq = seq.replace('N', 'A')
if ksize <= len(seq):
medn, ave, stdev = countgraph.get_median_count(seq)
ave, stdev = [round(x, 9) for x in (ave, stdev)]
output.writerow([record.name, medn, ave, stdev, len(seq)])
def test_abund_dist_gz_bigcount_compressed_first():
infile = utils.copy_test_data('test-abund-read-2.fa')
script = 'load-into-counting.py'
htfile = utils.get_temp_filename('test_ct.gz')
args = ['-x', str(1e7), '-N', str(2), '-k', str(2), htfile, infile]
utils.runscript(script, args) # create a bigcount table
assert os.path.exists(htfile)
data = gzip.open(htfile, 'rb').read() # read compressed bigcount table
outfile = utils.get_temp_filename('test_ct')
f_out = open(outfile, 'wb') # output the bigcount table
f_out.write(data)
f_out.close()
# load the compressed bigcount table
try:
countgraph = Countgraph.load(outfile)
except OSError as err:
assert 0, 'Should not produce OSError: ' + str(err)
assert countgraph.n_occupied() != 0
hashsizes = countgraph.hashsizes()
kmer_size = countgraph.ksize()
tracking = khmer.Nodegraph(kmer_size, 1, 1, primes=hashsizes)
abundances = countgraph.abundance_distribution(infile, tracking)
# calculate abundance distribution for compressed bigcount table
flag = False
# check if abundance is > 255
# if ok gzipped bigcount was loaded correctly
for _, i in enumerate(abundances):
print(_, i)
if _ > 255 and i > 0:
flag = True
break
assert flag
def test_load_gz():
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename('tempcountingsave1.ht')
loadpath = utils.get_temp_filename('tempcountingsave1.ht.gz')
sizes = list(PRIMES_1m)
sizes.append(1000005)
# save uncompressed hashtable.
hi = khmer.Countgraph(12, 1, 1, primes=sizes)
hi.consume_seqfile(inpath)
hi.save(savepath)
# compress.
in_file = open(savepath, 'rb')
out_file = gzip.open(loadpath, 'wb')
out_file.writelines(in_file)
out_file.close()
in_file.close()
# load compressed hashtable.
try:
ht = Countgraph.load(loadpath)
except OSError as err:
assert 0, "Should not produce an OSError: " + str(err)
tracking = khmer.Nodegraph(12, 1, 1, primes=sizes)
x = hi.abundance_distribution(inpath, tracking)
tracking = khmer.Nodegraph(12, 1, 1, primes=sizes)
y = ht.abundance_distribution(inpath, tracking)
assert sum(x) == 3966, sum(x)
assert x == y, (x, y)
def main():
counting_ht = sys.argv[1]
infiles = sys.argv[2:]
print('file with ht: %s' % counting_ht)
print('making hashtable')
ht = Countgraph.load(counting_ht)
K = ht.ksize()
for infile in infiles:
print('filtering', infile)
outfile = os.path.basename(infile) + '.below'
outfp = open(outfile, 'w')
paired_iter = broken_paired_reader(ReadParser(infile), min_length=K,
force_single=True)
for n, is_pair, read1, read2 in paired_iter:
name = read1.name
seq = read1.sequence
if 'N' in seq:
return None, None
trim_seq, trim_at = ht.trim_below_abundance(seq, CUTOFF)
if trim_at >= K:
write_record(screed.Record(name=name, sequence=trim_seq), outfp)
def test_load_into_counting_1():
in1 = utils.get_test_data('test-abund-read-2.fa')
out1 = utils.get_temp_filename('out.ct')
cmd = """
cat {in1} |
{scripts}/load-into-counting.py -x 1e3 -N 2 -k 20 {out1} - \
2> /dev/null
"""
cmd = cmd.format(scripts=scriptpath(), in1=in1, out1=out1)
print(cmd)
run_shell_cmd(cmd)
assert os.path.exists(out1)
Countgraph.load(out1)
def test_load_gz():
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename('tempcountingsave1.ht')
loadpath = utils.get_temp_filename('tempcountingsave1.ht.gz')
sizes = list(PRIMES_1m)
sizes.append(1000005)
# save uncompressed hashtable.
hi = khmer.Countgraph(12, 1, 1, primes=sizes)
hi.consume_seqfile(inpath)
hi.save(savepath)
# compress.
in_file = open(savepath, 'rb')
out_file = gzip.open(loadpath, 'wb')
out_file.writelines(in_file)
out_file.close()
in_file.close()
# load compressed hashtable.
try:
ht = Countgraph.load(loadpath)
except OSError as err:
assert 0, "Should not produce an OSError: " + str(err)
tracking = khmer.Nodegraph(12, 1, 1, primes=sizes)
x = hi.abundance_distribution(inpath, tracking)
tracking = khmer.Nodegraph(12, 1, 1, primes=sizes)
y = ht.abundance_distribution(inpath, tracking)
assert sum(x) == 3966, sum(x)
assert x == y, (x, y)
def test_badload():
try:
countgraph = Countgraph.load()
assert 0, "this should fail"
except TypeError as err:
print(str(err))
def test_abund_dist_gz_bigcount_compressed_first():
infile = utils.copy_test_data('test-abund-read-2.fa')
script = 'load-into-counting.py'
htfile = utils.get_temp_filename('test_ct.gz')
args = ['-x', str(1e7), '-N', str(2), '-k', str(2), htfile, infile]
utils.runscript(script, args) # create a bigcount table
assert os.path.exists(htfile)
data = gzip.open(htfile, 'rb').read() # read compressed bigcount table
outfile = utils.get_temp_filename('test_ct')
f_out = open(outfile, 'wb') # output the bigcount table
f_out.write(data)
f_out.close()
# load the compressed bigcount table
try:
countgraph = Countgraph.load(outfile)
except OSError as err:
assert 0, 'Should not produce OSError: ' + str(err)
assert countgraph.n_occupied() != 0
hashsizes = countgraph.hashsizes()
kmer_size = countgraph.ksize()
tracking = khmer.Nodegraph(kmer_size, 1, 1, primes=hashsizes)
abundances = countgraph.abundance_distribution(infile, tracking)
# calculate abundance distribution for compressed bigcount table
flag = False
# check if abundance is > 255
# if ok gzipped bigcount was loaded correctly
for _, i in enumerate(abundances):
print(_, i)
if _ > 255 and i > 0:
flag = True
break
assert flag
def test_badload():
try:
countgraph = Countgraph.load()
assert 0, "this should fail"
except TypeError as err:
print(str(err))
def test_load_into_counting_1():
in1 = utils.get_test_data('test-abund-read-2.fa')
out1 = utils.get_temp_filename('out.ct')
cmd = """
cat {in1} |
{scripts}/load-into-counting.py -x 1e3 -N 2 -k 20 {out1} - \
2> /dev/null
"""
cmd = cmd.format(scripts=scriptpath(), in1=in1, out1=out1)
print(cmd)
run_shell_cmd(cmd)
assert os.path.exists(out1)
Countgraph.load(out1)
def main():
info('count-kmers.py', ['counting'])
args = get_parser().parse_args()
print ('hashtable from', args.input_count_graph_filename,
file=sys.stderr)
countgraph = Countgraph.load(
args.input_count_graph_filename)
kmer_size = countgraph.ksize()
hashsizes = countgraph.hashsizes()
tracking = khmer.Nodegraph( # pylint: disable=protected-access
kmer_size, 1, 1, primes=hashsizes)
if args.output_file is None:
args.output_file = sys.stdout
writer = csv.writer(args.output_file)
for filename in args.input_sequence_filenames:
for record in screed.open(filename):
seq = record.sequence.replace('N', 'A')
for i in range(len(seq) - kmer_size + 1):
kmer = seq[i:i+kmer_size]
if not tracking.get(kmer):
tracking.count(kmer)
writer.writerow([kmer, str(countgraph.get(kmer))])
print ('Total number of unique k-mers: {0}'.format(
countgraph.n_unique_kmers()), file=sys.stderr)
def test_counting_gz_file_type_check():
inpath = utils.get_test_data('goodversion-k12.ht.gz')
try:
kh = Countgraph.load(inpath)
assert 0, "this should fail"
except OSError as e:
print(str(e))
def test_load_gz_notexist_should_fail():
savepath = utils.get_temp_filename('tempcountingsave0.ht.gz')
try:
hi = Countgraph.load(savepath)
assert 0, "load should fail"
except OSError as e:
print(str(e))
def test_load_notexist_should_fail():
savepath = utils.get_temp_filename('tempnodegraphsave0.htable')
try:
hi = Countgraph.load(savepath)
assert 0, "load should fail"
except OSError:
pass
def test_counting_gz_file_type_check():
inpath = utils.get_test_data('goodversion-k12.ht.gz')
try:
kh = Countgraph.load(inpath)
assert 0, "this should fail"
except OSError as e:
print(str(e))
def test_load_gz_notexist_should_fail():
savepath = utils.get_temp_filename('tempcountingsave0.ht.gz')
try:
hi = Countgraph.load(savepath)
assert 0, "load should fail"
except OSError as e:
print(str(e))
def test_load_notexist_should_fail():
savepath = utils.get_temp_filename('tempnodegraphsave0.htable')
try:
hi = Countgraph.load(savepath)
assert 0, "load should fail"
except OSError:
pass
def test_load_truncated():
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename('save.ht')
truncpath = utils.get_temp_filename('trunc.ht')
hi = khmer.Countgraph(12, 200, 3)
hi.consume_seqfile(inpath)
hi.save(savepath)
data = open(savepath, 'rb').read()
for i in range(len(data)):
fp = open(truncpath, 'wb')
fp.write(data[:i])
fp.close()
try:
Countgraph.load(truncpath)
assert 0, "this should not be reached!"
except OSError as err:
print(str(err))
def test_load_truncated():
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename('save.ht')
truncpath = utils.get_temp_filename('trunc.ht')
hi = khmer.Countgraph(12, 200, 3)
hi.consume_seqfile(inpath)
hi.save(savepath)
data = open(savepath, 'rb').read()
for i in range(len(data)):
fp = open(truncpath, 'wb')
fp.write(data[:i])
fp.close()
try:
Countgraph.load(truncpath)
assert 0, "this should not be reached!"
except OSError as err:
print(str(err))
def main():
parser = khmer_args.build_counting_args(
"Correct reads against an already-computed table",
citations=['counting', 'SeqAn'])
parser.add_argument("--trusted-cov",
dest="trusted_cov",
type=int,
default=DEFAULT_CUTOFF)
parser.add_argument("--theta", dest="bits_theta", type=float, default=1.0)
parser.add_argument('-o',
'--output',
dest='output_file',
help="output file for histogram; defaults to "
"<first filename>.corr in cwd.",
type=khFileType('w'),
default=None)
parser.add_argument('counts_table')
parser.add_argument('readfile')
args = parser.parse_args()
print('loading counts')
ht = Countgraph.load(args.counts_table)
aligner = khmer.ReadAligner(ht, args.trusted_cov, args.bits_theta)
print("trusted:", args.trusted_cov)
corrfp = args.output_file
if not corrfp:
outfile = os.path.basename(args.readfile) + '.corr'
corrfp = open(outfile, 'w')
n_corrected = 0
for n, read in enumerate(screed.open(args.readfile)):
if n % 10000 == 0:
print('...', n, n_corrected, file=sys.stderr)
seq = read.sequence.replace('N', 'A')
# build the alignment...
score, graph_alignment, read_alignment, truncated = \
aligner.align(seq)
if not truncated:
graph_seq = graph_alignment.replace("-", "")
if graph_seq != seq:
n_corrected += 1
seq = graph_seq
corrfp.write(output_single(read, seq))
def test_save_load_large(ctfile):
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename(ctfile)
orig = khmer.Countgraph(12, 2**31, 1)
orig.consume_seqfile(inpath)
orig.save(savepath)
loaded = Countgraph.load(savepath)
orig_count = orig.n_occupied()
loaded_count = loaded.n_occupied()
assert orig_count == 3966, orig_count
assert loaded_count == orig_count, loaded_count
def main():
parser = khmer_args.build_counting_args(
"Correct reads against an already-computed table",
citations=['counting', 'SeqAn'])
parser.add_argument("--trusted-cov", dest="trusted_cov", type=int,
default=DEFAULT_CUTOFF)
parser.add_argument("--theta", dest="bits_theta", type=float, default=1.0)
parser.add_argument('-o', '--output', dest='output_file',
help="output file for histogram; defaults to "
"<first filename>.corr in cwd.",
type=khFileType('w'), default=None)
parser.add_argument('counts_table')
parser.add_argument('readfile')
args = parser.parse_args()
print('loading counts')
ht = Countgraph.load(args.counts_table)
aligner = khmer.ReadAligner(ht,
args.trusted_cov,
args.bits_theta)
print("trusted:", args.trusted_cov)
corrfp = args.output_file
if not corrfp:
outfile = os.path.basename(args.readfile) + '.corr'
corrfp = open(outfile, 'w')
n_corrected = 0
for n, read in enumerate(screed.open(args.readfile)):
if n % 10000 == 0:
print('...', n, n_corrected, file=sys.stderr)
seq = read.sequence.replace('N', 'A')
# build the alignment...
score, graph_alignment, read_alignment, truncated = \
aligner.align(seq)
if not truncated:
graph_seq = graph_alignment.replace("-", "")
if graph_seq != seq:
n_corrected += 1
seq = graph_seq
corrfp.write(output_single(read, seq))
def test_save_load_large(ctfile):
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename(ctfile)
orig = khmer.Countgraph(12, 2**31, 1)
orig.consume_seqfile(inpath)
orig.save(savepath)
loaded = Countgraph.load(savepath)
orig_count = orig.n_occupied()
loaded_count = loaded.n_occupied()
assert orig_count == 3966, orig_count
assert loaded_count == orig_count, loaded_count
def test_save_load_occupied(ctfile):
print('working with', ctfile)
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename(ctfile)
orig = khmer.Countgraph(12, 1e5, 4)
orig.consume_seqfile(inpath)
orig.save(savepath)
loaded = Countgraph.load(savepath)
orig_count = orig.n_occupied()
loaded_count = loaded.n_occupied()
assert orig_count == 3886, orig_count
assert loaded_count == orig_count, loaded_count
def test_save_load_occupied(ctfile):
print('working with', ctfile)
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename(ctfile)
orig = khmer.Countgraph(12, 1e5, 4)
orig.consume_seqfile(inpath)
orig.save(savepath)
loaded = Countgraph.load(savepath)
orig_count = orig.n_occupied()
loaded_count = loaded.n_occupied()
assert orig_count == 3886, orig_count
assert loaded_count == orig_count, loaded_count
def test_maxcount_with_bigcount_save():
# hashtable should not saturate, if use_bigcount is set.
kh = khmer.Countgraph(4, 4 ** 4, 4)
kh.set_use_bigcount(True)
for _ in range(0, 1000):
kh.count('AAAA')
c = kh.get('AAAA')
savepath = utils.get_temp_filename('tempcountingsave.ht')
kh.save(savepath)
try:
kh = Countgraph.load(savepath)
except OSError as err:
assert 0, "Should not produce an OSError: " + str(err)
c = kh.get('AAAA')
assert c == 1000, "should be able to count to 1000: %d" % c
assert c != MAX_COUNT, c
def test_maxcount_with_bigcount_save():
# hashtable should not saturate, if use_bigcount is set.
kh = khmer.Countgraph(4, 4**4, 4)
kh.set_use_bigcount(True)
for _ in range(0, 1000):
kh.count('AAAA')
c = kh.get('AAAA')
savepath = utils.get_temp_filename('tempcountingsave.ht')
kh.save(savepath)
try:
kh = Countgraph.load(savepath)
except OSError as err:
assert 0, "Should not produce an OSError: " + str(err)
c = kh.get('AAAA')
assert c == 1000, "should be able to count to 1000: %d" % c
assert c != MAX_COUNT, c
def test_nobigcount_save():
kh = khmer.Countgraph(4, 4 ** 4, 4)
# kh.set_use_bigcount(False) <-- this is the default
savepath = utils.get_temp_filename('tempcountingsave.ht')
kh.save(savepath)
try:
kh = Countgraph.load(savepath)
except OSError as err:
assert 0, 'Should not produce an OSError: ' + str(err)
# set_use_bigcount should still be False after load (i.e. should be saved)
assert kh.get('AAAA') == 0
for _ in range(0, 1000):
kh.count('AAAA')
kh.get('AAAA')
assert kh.get('AAAA') == MAX_COUNT
def test_load_gz_truncated_should_fail():
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename('tempcountingsave0.ht.gz')
hi = khmer.Countgraph(12, 1000, 2)
hi.consume_seqfile(inpath)
hi.save(savepath)
fp = open(savepath, 'rb')
data = fp.read()
fp.close()
fp = open(savepath, 'wb')
fp.write(data[:1000])
fp.close()
try:
hi = Countgraph.load(savepath)
assert 0, "load should fail"
except OSError as e:
print(str(e))
def test_load_gz_truncated_should_fail():
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename('tempcountingsave0.ht.gz')
hi = khmer.Countgraph(12, 1000, 2)
hi.consume_seqfile(inpath)
hi.save(savepath)
fp = open(savepath, 'rb')
data = fp.read()
fp.close()
fp = open(savepath, 'wb')
fp.write(data[:1000])
fp.close()
try:
hi = Countgraph.load(savepath)
assert 0, "load should fail"
except OSError as e:
print(str(e))
def test_nobigcount_save():
kh = khmer.Countgraph(4, 4**4, 4)
# kh.set_use_bigcount(False) <-- this is the default
savepath = utils.get_temp_filename('tempcountingsave.ht')
kh.save(savepath)
try:
kh = Countgraph.load(savepath)
except OSError as err:
assert 0, 'Should not produce an OSError: ' + str(err)
# set_use_bigcount should still be False after load (i.e. should be saved)
assert kh.get('AAAA') == 0
for _ in range(0, 1000):
kh.count('AAAA')
kh.get('AAAA')
assert kh.get('AAAA') == MAX_COUNT
def test_bigcount_save():
# hashtable should not saturate, if use_bigcount is set.
kh = khmer.Countgraph(4, 4 ** 4, 4)
kh.set_use_bigcount(True)
savepath = utils.get_temp_filename('tempcountingsave.ht')
kh.save(savepath)
try:
kh = Countgraph.load(savepath)
except OSError as err:
assert 0, "Should not produce an OSError: " + str(err)
# set_use_bigcount should still be True after load (i.e. should be saved)
assert kh.get('AAAA') == 0
for _ in range(0, 1000):
kh.count('AAAA')
kh.get('AAAA')
assert kh.get('AAAA') == 1000
def test_bigcount_save():
# hashtable should not saturate, if use_bigcount is set.
kh = khmer.Countgraph(4, 4**4, 4)
kh.set_use_bigcount(True)
savepath = utils.get_temp_filename('tempcountingsave.ht')
kh.save(savepath)
try:
kh = Countgraph.load(savepath)
except OSError as err:
assert 0, "Should not produce an OSError: " + str(err)
# set_use_bigcount should still be True after load (i.e. should be saved)
assert kh.get('AAAA') == 0
for _ in range(0, 1000):
kh.count('AAAA')
kh.get('AAAA')
assert kh.get('AAAA') == 1000
def test_save_load_gz():
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename('tempcountingsave2.ht.gz')
sizes = list(PRIMES_1m)
sizes.append(1000005)
hi = khmer.Countgraph(12, 1, 1, primes=sizes)
hi.consume_seqfile(inpath)
hi.save(savepath)
try:
ht = Countgraph.load(savepath)
except OSError as err:
assert 0, 'Should not produce an OSError: ' + str(err)
tracking = khmer.Nodegraph(12, 1, 1, primes=sizes)
x = hi.abundance_distribution(inpath, tracking)
tracking = khmer.Nodegraph(12, 1, 1, primes=sizes)
y = ht.abundance_distribution(inpath, tracking)
assert sum(x) == 3966, sum(x)
assert x == y, (x, y)
def test_save_load_gz():
inpath = utils.get_test_data('random-20-a.fa')
savepath = utils.get_temp_filename('tempcountingsave2.ht.gz')
sizes = list(PRIMES_1m)
sizes.append(1000005)
hi = khmer.Countgraph(12, 1, 1, primes=sizes)
hi.consume_seqfile(inpath)
hi.save(savepath)
try:
ht = Countgraph.load(savepath)
except OSError as err:
assert 0, 'Should not produce an OSError: ' + str(err)
tracking = khmer.Nodegraph(12, 1, 1, primes=sizes)
x = hi.abundance_distribution(inpath, tracking)
tracking = khmer.Nodegraph(12, 1, 1, primes=sizes)
y = ht.abundance_distribution(inpath, tracking)
assert sum(x) == 3966, sum(x)
assert x == y, (x, y)
def main():
info('correct-reads.py', ['streaming'])
args = sanitize_help(get_parser()).parse_args()
###
if len(set(args.input_filenames)) != len(args.input_filenames):
print("Error: Cannot input the same filename multiple times.",
file=sys.stderr)
sys.exit(1)
###
report_on_config(args)
check_valid_file_exists(args.input_filenames)
check_space(args.input_filenames, args.force)
tablesize = calculate_graphsize(args, 'countgraph')
if args.savegraph:
check_space_for_graph(args.savegraph, tablesize, args.force)
K = args.ksize
CUTOFF = args.cutoff
NORMALIZE_LIMIT = args.normalize_to
if args.loadgraph:
print('loading k-mer countgraph from', args.loadgraph, file=sys.stderr)
ct = Countgraph.load(args.loadgraph)
else:
print('making k-mer countgraph', file=sys.stderr)
ct = create_countgraph(args, multiplier=8 / (9. + 0.3))
tempdir = tempfile.mkdtemp('khmer', 'tmp', args.tempdir)
print('created temporary directory %s; use -T to change location' %
tempdir,
file=sys.stderr)
aligner = khmer.ReadAligner(ct, args.cutoff, args.bits_theta)
# ### FIRST PASS ###
save_pass2_total = 0
n_bp = 0
n_reads = 0
written_bp = 0
written_reads = 0
corrected_reads = 0
pass2list = []
for filename in args.input_filenames:
pass2filename = os.path.basename(filename) + '.pass2'
pass2filename = os.path.join(tempdir, pass2filename)
if args.out is None:
corrfp = open(os.path.basename(filename) + '.corr', 'w')
else:
corrfp = args.out
pass2list.append((filename, pass2filename, corrfp))
screed_iter = screed.open(filename, parse_description=False)
pass2fp = open(pass2filename, 'w')
save_pass2 = 0
n = 0
paired_iter = broken_paired_reader(screed_iter,
min_length=K,
force_single=args.ignore_pairs)
for n, is_pair, read1, read2 in paired_iter:
if n % 10000 == 0:
print('...',
n,
filename,
save_pass2,
n_reads,
n_bp,
written_reads,
written_bp,
file=sys.stderr)
# we want to track paired reads here, to make sure that pairs
# are not split between first pass and second pass.
if is_pair:
n_reads += 2
n_bp += len(read1.sequence) + len(read2.sequence)
seq1 = read1.sequence.replace('N', 'A')
seq2 = read2.sequence.replace('N', 'A')
med1, _, _ = ct.get_median_count(seq1)
med2, _, _ = ct.get_median_count(seq2)
if med1 < NORMALIZE_LIMIT or med2 < NORMALIZE_LIMIT:
ct.consume(seq1)
ct.consume(seq2)
write_record_pair(read1, read2, pass2fp)
save_pass2 += 2
else:
is_aligned, new_seq1 = correct_sequence(aligner, seq1)
if is_aligned:
if new_seq1 != read1.sequence:
corrected_reads += 1
read1.sequence = new_seq1
if hasattr(read1, 'quality'):
fix_quality(read1)
is_aligned, new_seq2 = correct_sequence(aligner, seq2)
if is_aligned:
if new_seq2 != read2.sequence:
corrected_reads += 1
read2.sequence = new_seq2
if hasattr(read2, 'quality'):
fix_quality(read2)
write_record_pair(read1, read2, corrfp)
written_reads += 2
written_bp += len(read1)
written_bp += len(read2)
else:
n_reads += 1
n_bp += len(read1.sequence)
seq = read1.sequence.replace('N', 'A')
med, _, _ = ct.get_median_count(seq)
# has this portion of the graph saturated? if not,
# consume & save => pass2.
if med < NORMALIZE_LIMIT:
ct.consume(seq)
write_record(read1, pass2fp)
save_pass2 += 1
else: # trim!!
is_aligned, new_seq = correct_sequence(aligner, seq)
if is_aligned:
if new_seq != read1.sequence:
corrected_reads += 1
read1.sequence = new_seq
if hasattr(read1, 'quality'):
fix_quality(read1)
write_record(read1, corrfp)
written_reads += 1
written_bp += len(new_seq)
pass2fp.close()
print('%s: kept aside %d of %d from first pass, in %s' %
(filename, save_pass2, n, filename),
file=sys.stderr)
save_pass2_total += save_pass2
# ### SECOND PASS. ###
skipped_n = 0
skipped_bp = 0
for _, pass2filename, corrfp in pass2list:
print(('second pass: looking at sequences kept aside in %s') %
pass2filename,
file=sys.stderr)
# note that for this second pass, we don't care about paired
# reads - they will be output in the same order they're read in,
# so pairs will stay together if not orphaned. This is in contrast
# to the first loop.
for n, read in enumerate(
screed.open(pass2filename, parse_description=False)):
if n % 10000 == 0:
print('... x 2',
n,
pass2filename,
written_reads,
written_bp,
file=sys.stderr)
seq = read.sequence.replace('N', 'A')
med, _, _ = ct.get_median_count(seq)
# do we retain low-abundance components unchanged?
if med < NORMALIZE_LIMIT and args.variable_coverage:
write_record(read, corrfp)
written_reads += 1
written_bp += len(read.sequence)
skipped_n += 1
skipped_bp += len(read.sequence)
# otherwise, examine/correct.
else: # med >= NORMALIZE LIMIT or not args.variable_coverage
is_aligned, new_seq = correct_sequence(aligner, seq)
if is_aligned:
if new_seq != read.sequence:
corrected_reads += 1
read.sequence = new_seq
if hasattr(read, 'quality'):
fix_quality(read)
write_record(read, corrfp)
written_reads += 1
written_bp += len(new_seq)
print('removing %s' % pass2filename, file=sys.stderr)
os.unlink(pass2filename)
print('removing temp directory & contents (%s)' % tempdir, file=sys.stderr)
shutil.rmtree(tempdir)
n_passes = 1.0 + (float(save_pass2_total) / n_reads)
percent_reads_corrected = float(corrected_reads +
(n_reads - written_reads)) /\
n_reads * 100.0
print('read %d reads, %d bp' % (
n_reads,
n_bp,
), file=sys.stderr)
print('wrote %d reads, %d bp' % (
written_reads,
written_bp,
),
file=sys.stderr)
print('looked at %d reads twice (%.2f passes)' %
(save_pass2_total, n_passes),
file=sys.stderr)
print('removed %d reads and corrected %d reads (%.2f%%)' %
(n_reads - written_reads, corrected_reads, percent_reads_corrected),
file=sys.stderr)
print('removed %.2f%% of bases (%d total)' %
((1 - (written_bp / float(n_bp))) * 100.0, n_bp - written_bp),
file=sys.stderr)
if args.variable_coverage:
percent_reads_hicov = 100.0 * float(n_reads - skipped_n) / n_reads
print('%d reads were high coverage (%.2f%%);' %
(n_reads - skipped_n, percent_reads_hicov),
file=sys.stderr)
print(('skipped %d reads/%d bases because of low coverage') %
(skipped_n, skipped_bp),
file=sys.stderr)
fp_rate = \
khmer.calc_expected_collisions(ct, args.force, max_false_pos=.8)
# for max_false_pos see Zhang et al., http://arxiv.org/abs/1309.2975
print('fp rate estimated to be {fpr:1.3f}'.format(fpr=fp_rate),
file=sys.stderr)
print('output in *.corr', file=sys.stderr)
if args.savegraph:
print("Saving k-mer countgraph to", args.savegraph, file=sys.stderr)
ct.save(args.savegraph)
def main():
parser = sanitize_help(get_parser())
args = parser.parse_args()
configure_logging(args.quiet)
###
if len(set(args.input_filenames)) != len(args.input_filenames):
log_error("Error: Cannot input the same filename multiple times.")
sys.exit(1)
if args.trim_at_coverage != DEFAULT_TRIM_AT_COVERAGE and \
not args.variable_coverage:
log_error("Error: --trim-at-coverage/-Z given, but "
"--variable-coverage/-V not specified.")
sys.exit(1)
if args.diginorm_coverage != DEFAULT_DIGINORM_COVERAGE and \
not args.diginorm:
log_error("Error: --diginorm-coverage given, but "
"--diginorm not specified.")
sys.exit(1)
if args.diginorm and args.single_pass:
log_error("Error: --diginorm and --single-pass are incompatible!\n"
"You probably want to use normalize-by-median.py instead.")
sys.exit(1)
###
graphtype = 'countgraph' if not args.small_count else 'smallcountgraph'
report_on_config(args, graphtype=graphtype)
check_valid_file_exists(args.input_filenames)
check_space(args.input_filenames, args.force)
if args.savegraph:
graphsize = calculate_graphsize(args, graphtype)
check_space_for_graph(args.savegraph, graphsize, args.force)
if ('-' in args.input_filenames or '/dev/stdin' in args.input_filenames) \
and not args.output:
log_error("Accepting input from stdin; output filename must "
"be provided with -o.")
sys.exit(1)
if args.loadgraph:
log_info('loading countgraph from {graph}', graph=args.loadgraph)
if args.small_count:
ct = SmallCountgraph.load(args.loadgraph)
else:
ct = Countgraph.load(args.loadgraph)
else:
log_info('making countgraph')
ct = khmer_args.create_countgraph(args)
K = ct.ksize()
tempdir = tempfile.mkdtemp('khmer', 'tmp', args.tempdir)
log_info(
'created temporary directory {temp};\n'
'use -T to change location',
temp=tempdir)
trimmer = Trimmer(ct, not args.variable_coverage, args.cutoff,
args.trim_at_coverage)
if args.diginorm:
trimmer.set_diginorm(args.diginorm_coverage)
# ### FIRST PASS ###
save_pass2_total = 0
written_bp = 0
written_reads = 0
# only create the file writer once if outfp is specified; otherwise,
# create it for each file.
if args.output:
trimfp = get_file_writer(args.output, args.gzip, args.bzip)
pass2list = []
for filename in args.input_filenames:
# figure out temporary filename for 2nd pass
pass2filename = os.path.basename(filename) + '.pass2'
pass2filename = os.path.join(tempdir, pass2filename)
pass2fp = open(pass2filename, 'w')
# construct output filenames
if args.output is None:
# note: this will be saved in trimfp.
outfp = open(os.path.basename(filename) + '.abundtrim', 'wb')
# get file handle w/gzip, bzip
trimfp = get_file_writer(outfp, args.gzip, args.bzip)
# record all this info
pass2list.append((filename, pass2filename, trimfp))
# input file stuff: get a broken_paired reader.
paired_iter = broken_paired_reader(ReadParser(filename),
min_length=K,
force_single=args.ignore_pairs)
# main loop through the file.
n_start = trimmer.n_reads
save_start = trimmer.n_saved
watermark = REPORT_EVERY_N_READS
for read in trimmer.pass1(paired_iter, pass2fp):
if (trimmer.n_reads - n_start) > watermark:
log_info(
"... {filename} {n_saved} {n_reads} {n_bp} "
"{w_reads} {w_bp}",
filename=filename,
n_saved=trimmer.n_saved,
n_reads=trimmer.n_reads,
n_bp=trimmer.n_bp,
w_reads=written_reads,
w_bp=written_bp)
watermark += REPORT_EVERY_N_READS
# write out the trimmed/etc sequences that AREN'T going to be
# revisited in a 2nd pass.
write_record(read, trimfp)
written_bp += len(read)
written_reads += 1
pass2fp.close()
log_info("{filename}: kept aside {kept} of {total} from first pass",
filename=filename,
kept=trimmer.n_saved - save_start,
total=trimmer.n_reads - n_start)
# first pass goes across all the data, so record relevant stats...
n_reads = trimmer.n_reads
n_bp = trimmer.n_bp
n_skipped = trimmer.n_skipped
bp_skipped = trimmer.bp_skipped
save_pass2_total = trimmer.n_saved
# ### SECOND PASS. ###
# nothing should have been skipped yet!
assert trimmer.n_skipped == 0
assert trimmer.bp_skipped == 0
if args.single_pass:
pass2list = []
# go back through all the files again.
for _, pass2filename, trimfp in pass2list:
log_info('second pass: looking at sequences kept aside in {pass2}',
pass2=pass2filename)
# note that for this second pass, we don't care about paired
# reads - they will be output in the same order they're read in,
# so pairs will stay together if not orphaned. This is in contrast
# to the first loop. Hence, force_single=True below.
read_parser = ReadParser(pass2filename)
paired_iter = broken_paired_reader(read_parser,
min_length=K,
force_single=True)
watermark = REPORT_EVERY_N_READS
for read in trimmer.pass2(paired_iter):
if (trimmer.n_reads - n_start) > watermark:
log_info('... x 2 {a} {b} {c} {d} {e} {f} {g}',
a=trimmer.n_reads - n_start,
b=pass2filename,
c=trimmer.n_saved,
d=trimmer.n_reads,
e=trimmer.n_bp,
f=written_reads,
g=written_bp)
watermark += REPORT_EVERY_N_READS
write_record(read, trimfp)
written_reads += 1
written_bp += len(read)
read_parser.close()
log_info('removing {pass2}', pass2=pass2filename)
os.unlink(pass2filename)
# if we created our own trimfps, close 'em.
if not args.output:
trimfp.close()
try:
log_info('removing temp directory & contents ({temp})', temp=tempdir)
shutil.rmtree(tempdir)
except OSError as oe:
log_info(
'WARNING: unable to remove {temp} (probably an NFS issue); '
'please remove manually',
temp=tempdir)
trimmed_reads = trimmer.trimmed_reads
n_passes = 1.0 + (float(save_pass2_total) / n_reads)
percent_reads_trimmed = float(trimmed_reads + (n_reads - written_reads)) /\
n_reads * 100.0
log_info('read {read} reads, {bp} bp', read=n_reads, bp=n_bp)
log_info('wrote {wr} reads, {wbp} bp', wr=written_reads, wbp=written_bp)
log_info('looked at {st} reads twice ({np:.2f} passes)',
st=save_pass2_total,
np=n_passes)
log_info('removed {r} reads and trimmed {t} reads ({p:.2f}%)',
r=n_reads - written_reads,
t=trimmed_reads,
p=percent_reads_trimmed)
log_info('trimmed or removed {p:.2f}%% of bases ({bp} total)',
p=(1 - (written_bp / float(n_bp))) * 100.0,
bp=n_bp - written_bp)
if args.variable_coverage:
percent_reads_hicov = 100.0 * float(n_reads - n_skipped) / n_reads
log_info('{n} reads were high coverage ({p:.2f}%);',
n=n_reads - n_skipped,
p=percent_reads_hicov)
log_info('skipped {r} reads/{bp} bases because of low coverage',
r=n_skipped,
bp=bp_skipped)
fp_rate = \
khmer.calc_expected_collisions(ct, args.force, max_false_pos=.8)
# for max_false_pos see Zhang et al., http://arxiv.org/abs/1309.2975
log_info('fp rate estimated to be {fpr:1.3f}', fpr=fp_rate)
if args.output is None:
log_info('output in *.abundtrim')
elif args.output.name == 1:
log_info('output streamed to stdout')
elif args.output.name:
log_info('output in {}'.format(args.output.name))
if args.savegraph:
log_info("Saving k-mer countgraph to {graph}", graph=args.savegraph)
ct.save(args.savegraph)
if args.summary_info is not None:
# note that when streaming to stdout the name of args.output will
# be set to 1
if args.output is not None and args.output.name != 1:
base = args.output.name
# no explicit name or stdout stream -> use a default name
else:
base = 'trim-low-abund-{}'.format(
time.strftime("%Y-%m-%dT%H:%M:%S"))
info = {
'fpr': fp_rate,
'reads': n_reads,
'basepairs': n_bp,
'reads_written': written_reads,
'basepairs_written': written_bp,
'reads_skipped': n_skipped,
'basepairs_skipped': bp_skipped,
'reads_removed': n_reads - written_reads,
'reads_trimmed': trimmed_reads,
'basepairs_removed_or_trimmed': n_bp - written_bp
}
store_provenance_info(info, fname=base, format=args.summary_info)
def _get_contaminants(self):
from khmer import Countgraph, khmer_args
# assuming all sequences are same length
n_win = self._read_length - self.kmer_size + 1
tablesize = self.n_reads * n_win
countgraph = Countgraph(
self.kmer_size, tablesize, khmer_args.DEFAULT_N_TABLES)
countgraph.set_use_bigcount(True)
for seq in self._read_sequences:
countgraph.consume_and_tag(seq)
n_expected = math.ceil(tablesize / float(4**self.kmer_size))
min_count = n_expected * self.overrep_cutoff
if min_count >= 2**16:
raise ValueError(
"The minimum count for an over-represented k-kmer {} is "
"greater than the max khmer count (2^16)".format(min_count))
candidates = {}
for tag in countgraph.get_tagset():
count = countgraph.get(tag)
if count >= min_count:
candidates[tag] = count
if self.known_contaminants:
matches = []
seen = set()
def match(kmer):
"""Returns the frequency of `kmer` in `candidates`.
"""
freq = candidates.get(kmer, 0)
if freq > 0:
seen.add(kmer)
return freq
for seq, names in self.known_contaminants.iter_sequences():
seqlen = len(seq)
if seqlen < self.kmer_size:
print("Cannot check {}; sequence is shorter than {}".format(
list(names)[0], self.kmer_size))
continue
n_kmers = seqlen - self.kmer_size + 1
num_matches = 0
match_counts = []
for idx in range(n_kmers):
kmer = seq[idx:(idx + self.kmer_size)]
kmer_count = max(
match(kmer),
match(reverse_complement(kmer))
)
if kmer_count > 0:
num_matches += 1
match_counts.append(kmer_count)
if num_matches > 0:
# not sure what the correct metric is to use here
overall_count = sum(match_counts) / float(n_kmers)
matches.append(Match(
seq, count=overall_count / float(tablesize),
names=names, match_frac=float(num_matches) / n_kmers))
# Add remaining tags
for tag in set(candidates.keys()) - seen:
matches.append(Match(
tag, count=candidates[tag] / float(tablesize)))
else:
matches = [
Match(tag, count=count / float(tablesize))
for tag, count in candidates.items()]
return matches
def main():
parser = sanitize_help(get_parser())
args = parser.parse_args()
configure_logging(args.quiet)
###
if len(set(args.input_filenames)) != len(args.input_filenames):
log_error("Error: Cannot input the same filename multiple times.")
sys.exit(1)
if args.trim_at_coverage != DEFAULT_TRIM_AT_COVERAGE and \
not args.variable_coverage:
log_error("Error: --trim-at-coverage/-Z given, but "
"--variable-coverage/-V not specified.")
sys.exit(1)
if args.diginorm_coverage != DEFAULT_DIGINORM_COVERAGE and \
not args.diginorm:
log_error("Error: --diginorm-coverage given, but "
"--diginorm not specified.")
sys.exit(1)
if args.diginorm and args.single_pass:
log_error("Error: --diginorm and --single-pass are incompatible!\n"
"You probably want to use normalize-by-median.py instead.")
sys.exit(1)
###
graphtype = 'countgraph' if not args.small_count else 'smallcountgraph'
report_on_config(args, graphtype=graphtype)
check_valid_file_exists(args.input_filenames)
check_space(args.input_filenames, args.force)
if args.savegraph:
graphsize = calculate_graphsize(args, graphtype)
check_space_for_graph(args.savegraph, graphsize, args.force)
if ('-' in args.input_filenames or '/dev/stdin' in args.input_filenames) \
and not args.output:
log_error("Accepting input from stdin; output filename must "
"be provided with -o.")
sys.exit(1)
if args.loadgraph:
log_info('loading countgraph from {graph}', graph=args.loadgraph)
if args.small_count:
ct = SmallCountgraph.load(args.loadgraph)
else:
ct = Countgraph.load(args.loadgraph)
else:
log_info('making countgraph')
ct = khmer_args.create_countgraph(args)
K = ct.ksize()
tempdir = tempfile.mkdtemp('khmer', 'tmp', args.tempdir)
log_info('created temporary directory {temp};\n'
'use -T to change location', temp=tempdir)
trimmer = Trimmer(ct, not args.variable_coverage, args.cutoff,
args.trim_at_coverage)
if args.diginorm:
trimmer.set_diginorm(args.diginorm_coverage)
# ### FIRST PASS ###
save_pass2_total = 0
written_bp = 0
written_reads = 0
# only create the file writer once if outfp is specified; otherwise,
# create it for each file.
if args.output:
trimfp = get_file_writer(args.output, args.gzip, args.bzip)
pass2list = []
for filename in args.input_filenames:
# figure out temporary filename for 2nd pass
pass2filename = os.path.basename(filename) + '.pass2'
pass2filename = os.path.join(tempdir, pass2filename)
pass2fp = open(pass2filename, 'w')
# construct output filenames
if args.output is None:
# note: this will be saved in trimfp.
outfp = open(os.path.basename(filename) + '.abundtrim', 'wb')
# get file handle w/gzip, bzip
trimfp = get_file_writer(outfp, args.gzip, args.bzip)
# record all this info
pass2list.append((filename, pass2filename, trimfp))
# input file stuff: get a broken_paired reader.
paired_iter = broken_paired_reader(ReadParser(filename), min_length=K,
force_single=args.ignore_pairs)
# main loop through the file.
n_start = trimmer.n_reads
save_start = trimmer.n_saved
watermark = REPORT_EVERY_N_READS
for read in trimmer.pass1(paired_iter, pass2fp):
if (trimmer.n_reads - n_start) > watermark:
log_info("... {filename} {n_saved} {n_reads} {n_bp} "
"{w_reads} {w_bp}", filename=filename,
n_saved=trimmer.n_saved, n_reads=trimmer.n_reads,
n_bp=trimmer.n_bp, w_reads=written_reads,
w_bp=written_bp)
watermark += REPORT_EVERY_N_READS
# write out the trimmed/etc sequences that AREN'T going to be
# revisited in a 2nd pass.
write_record(read, trimfp)
written_bp += len(read)
written_reads += 1
pass2fp.close()
log_info("{filename}: kept aside {kept} of {total} from first pass",
filename=filename, kept=trimmer.n_saved - save_start,
total=trimmer.n_reads - n_start)
# first pass goes across all the data, so record relevant stats...
n_reads = trimmer.n_reads
n_bp = trimmer.n_bp
n_skipped = trimmer.n_skipped
bp_skipped = trimmer.bp_skipped
save_pass2_total = trimmer.n_saved
# ### SECOND PASS. ###
# nothing should have been skipped yet!
assert trimmer.n_skipped == 0
assert trimmer.bp_skipped == 0
if args.single_pass:
pass2list = []
# go back through all the files again.
for _, pass2filename, trimfp in pass2list:
log_info('second pass: looking at sequences kept aside in {pass2}',
pass2=pass2filename)
# note that for this second pass, we don't care about paired
# reads - they will be output in the same order they're read in,
# so pairs will stay together if not orphaned. This is in contrast
# to the first loop. Hence, force_single=True below.
read_parser = ReadParser(pass2filename)
paired_iter = broken_paired_reader(read_parser,
min_length=K,
force_single=True)
watermark = REPORT_EVERY_N_READS
for read in trimmer.pass2(paired_iter):
if (trimmer.n_reads - n_start) > watermark:
log_info('... x 2 {a} {b} {c} {d} {e} {f} {g}',
a=trimmer.n_reads - n_start,
b=pass2filename, c=trimmer.n_saved,
d=trimmer.n_reads, e=trimmer.n_bp,
f=written_reads, g=written_bp)
watermark += REPORT_EVERY_N_READS
write_record(read, trimfp)
written_reads += 1
written_bp += len(read)
read_parser.close()
log_info('removing {pass2}', pass2=pass2filename)
os.unlink(pass2filename)
# if we created our own trimfps, close 'em.
if not args.output:
trimfp.close()
try:
log_info('removing temp directory & contents ({temp})', temp=tempdir)
shutil.rmtree(tempdir)
except OSError as oe:
log_info('WARNING: unable to remove {temp} (probably an NFS issue); '
'please remove manually', temp=tempdir)
trimmed_reads = trimmer.trimmed_reads
n_passes = 1.0 + (float(save_pass2_total) / n_reads)
percent_reads_trimmed = float(trimmed_reads + (n_reads - written_reads)) /\
n_reads * 100.0
log_info('read {read} reads, {bp} bp', read=n_reads, bp=n_bp)
log_info('wrote {wr} reads, {wbp} bp', wr=written_reads, wbp=written_bp)
log_info('looked at {st} reads twice ({np:.2f} passes)',
st=save_pass2_total, np=n_passes)
log_info('removed {r} reads and trimmed {t} reads ({p:.2f}%)',
r=n_reads - written_reads, t=trimmed_reads,
p=percent_reads_trimmed)
log_info('trimmed or removed {p:.2f}%% of bases ({bp} total)',
p=(1 - (written_bp / float(n_bp))) * 100.0, bp=n_bp - written_bp)
if args.variable_coverage:
percent_reads_hicov = 100.0 * float(n_reads - n_skipped) / n_reads
log_info('{n} reads were high coverage ({p:.2f}%);',
n=n_reads - n_skipped, p=percent_reads_hicov)
log_info('skipped {r} reads/{bp} bases because of low coverage',
r=n_skipped, bp=bp_skipped)
fp_rate = \
khmer.calc_expected_collisions(ct, args.force, max_false_pos=.8)
# for max_false_pos see Zhang et al., http://arxiv.org/abs/1309.2975
log_info('fp rate estimated to be {fpr:1.3f}', fpr=fp_rate)
if args.output is None:
log_info('output in *.abundtrim')
elif args.output.name == 1:
log_info('output streamed to stdout')
elif args.output.name:
log_info('output in {}'.format(args.output.name))
if args.savegraph:
log_info("Saving k-mer countgraph to {graph}", graph=args.savegraph)
ct.save(args.savegraph)
if args.summary_info is not None:
# note that when streaming to stdout the name of args.output will
# be set to 1
if args.output is not None and args.output.name != 1:
base = args.output.name
# no explicit name or stdout stream -> use a default name
else:
base = 'trim-low-abund-{}'.format(
time.strftime("%Y-%m-%dT%H:%M:%S"))
info = {'fpr': fp_rate,
'reads': n_reads,
'basepairs': n_bp,
'reads_written': written_reads,
'basepairs_written': written_bp,
'reads_skipped': n_skipped,
'basepairs_skipped': bp_skipped,
'reads_removed': n_reads - written_reads,
'reads_trimmed': trimmed_reads,
'basepairs_removed_or_trimmed': n_bp - written_bp
}
store_provenance_info(info, fname=base, format=args.summary_info)
def main():
args = sanitize_help(get_parser()).parse_args()
configure_logging(args.quiet)
infiles = [args.input_count_graph_filename, args.input_sequence_filename]
for infile in infiles:
check_input_files(infile, False)
log_info('Loading counting graph from {graph}',
graph=args.input_count_graph_filename)
countgraph = Countgraph.load(args.input_count_graph_filename)
if not countgraph.get_use_bigcount() and args.bigcount:
log_warn("WARNING: The loaded graph has bigcount DISABLED while "
"bigcount reporting is ENABLED--counts higher than 255 will "
"not be reported.")
countgraph.set_use_bigcount(args.bigcount)
kmer_size = countgraph.ksize()
hashsizes = countgraph.hashsizes()
tracking = khmer.Nodegraph( # pylint: disable=protected-access
kmer_size, 1, 1, primes=hashsizes)
log_info('K: {ksize}', ksize=kmer_size)
log_info('outputting to {output}', output=args.output_histogram_filename)
if args.output_histogram_filename in ('-', '/dev/stdout'):
pass
elif os.path.exists(args.output_histogram_filename):
if not args.squash_output:
log_error('ERROR: {output} exists; not squashing.',
output=args.output_histogram_filename)
sys.exit(1)
log_info('** squashing existing file {output}',
output=args.output_histogram_filename)
log_info('preparing hist...')
abundances = countgraph.abundance_distribution(
args.input_sequence_filename, tracking)
total = sum(abundances)
if 0 == total:
log_error("ERROR: abundance distribution is uniformly zero; "
"nothing to report.")
log_error("\tPlease verify that the input files are valid.")
sys.exit(1)
if args.output_histogram_filename in ('-', '/dev/stdout'):
countgraph_fp = sys.stdout
else:
countgraph_fp = open(args.output_histogram_filename, 'w')
countgraph_fp_csv = csv.writer(countgraph_fp)
# write headers:
countgraph_fp_csv.writerow(
['abundance', 'count', 'cumulative', 'cumulative_fraction'])
sofar = 0
for _, i in enumerate(abundances):
if i == 0 and not args.output_zero:
continue
sofar += i
frac = sofar / float(total)
countgraph_fp_csv.writerow([_, i, sofar, round(frac, 3)])
if sofar == total:
break
def test_load_empty_files(ext):
# Check empty files, compressed or not
fname = utils.get_test_data('empty-file' + ext)
with pytest.raises(OSError):
Countgraph.load(fname)
def main(): # pylint: disable=too-many-branches,too-many-statements
parser = sanitize_help(get_parser())
args = parser.parse_args()
configure_logging(args.quiet)
report_on_config(args)
report_fp = args.report
force_single = args.force_single
# check for similar filenames
# if we're using a single output file only check for identical filenames
# otherwise, check for identical BASE names as well.
filenames = []
basenames = []
for pathfilename in args.input_filenames:
filenames.append(pathfilename)
if args.single_output_file:
continue # nothing more to worry about
basename = os.path.basename(pathfilename)
if basename in basenames:
log_error('ERROR: Duplicate filename--Cannot handle this!')
log_error('** Exiting!')
sys.exit(1)
basenames.append(basename)
# check that files exist and there is sufficient output disk space.
check_valid_file_exists(args.input_filenames)
check_space(args.input_filenames, args.force)
if args.savegraph is not None:
graphsize = calculate_graphsize(args, 'countgraph')
check_space_for_graph(args.savegraph, graphsize, args.force)
# load or create counting table.
if args.loadgraph:
log_info('loading k-mer countgraph from {graph}',
graph=args.loadgraph)
countgraph = Countgraph.load(args.loadgraph)
else:
log_info('making countgraph')
countgraph = khmer_args.create_countgraph(args)
# create an object to handle diginorm of all files
norm = Normalizer(args.cutoff, countgraph)
with_diagnostics = WithDiagnostics(norm, report_fp, args.report_frequency)
# make a list of all filenames and if they're paired or not;
# if we don't know if they're paired, default to allowing but not
# forcing pairing.
files = []
for element in filenames:
files.append([element, args.paired])
if args.unpaired_reads:
files.append([args.unpaired_reads, False])
corrupt_files = []
outfp = None
output_name = None
if args.single_output_file:
outfp = get_file_writer(args.single_output_file, args.gzip, args.bzip)
else:
if '-' in filenames or '/dev/stdin' in filenames:
print("Accepting input from stdin; output filename must "
"be provided with '-o'.", file=sys.stderr)
sys.exit(1)
#
# main loop: iterate over all files given, do diginorm.
#
for filename, require_paired in files:
if not args.single_output_file:
output_name = os.path.basename(filename) + '.keep'
outfp = open(output_name, 'wb')
outfp = get_file_writer(outfp, args.gzip, args.bzip)
# failsafe context manager in case an input file breaks
with catch_io_errors(filename, outfp, args.single_output_file,
args.force, corrupt_files):
screed_iter = clean_input_reads(screed.open(filename))
reader = broken_paired_reader(screed_iter, min_length=args.ksize,
force_single=force_single,
require_paired=require_paired)
# actually do diginorm
for record in with_diagnostics(reader, filename):
if record is not None:
write_record(record, outfp)
log_info('output in {name}', name=describe_file_handle(outfp))
if not args.single_output_file:
outfp.close()
# finished - print out some diagnostics.
log_info('Total number of unique k-mers: {umers}',
umers=countgraph.n_unique_kmers())
if args.savegraph is not None:
log_info('...saving to {name}', name=args.savegraph)
countgraph.save(args.savegraph)
fp_rate = \
khmer.calc_expected_collisions(countgraph, False, max_false_pos=.8)
# for max_false_pos see Zhang et al., http://arxiv.org/abs/1309.2975
log_info('fp rate estimated to be {fpr:1.3f}', fpr=fp_rate)
if args.force and len(corrupt_files) > 0:
log_error("** WARNING: Finished with errors!")
log_error("** I/O Errors occurred in the following files:")
log_error("\t" + " ".join(corrupt_files))
def test_load_empty_files(ext):
# Check empty files, compressed or not
fname = utils.get_test_data('empty-file' + ext)
with pytest.raises(OSError):
Countgraph.load(fname)
def main(): # pylint: disable=too-many-branches,too-many-statements
start_time = time.time()
parser = sanitize_help(get_parser())
args = parser.parse_args()
configure_logging(args.quiet)
report_on_config(args)
report_fp = args.report
force_single = args.force_single
# check for similar filenames
# if we're using a single output file only check for identical filenames
# otherwise, check for identical BASE names as well.
filenames = []
basenames = []
for pathfilename in args.input_filenames:
filenames.append(pathfilename)
if args.single_output_file:
continue # nothing more to worry about
basename = os.path.basename(pathfilename)
if basename in basenames:
log_error('ERROR: Duplicate filename--Cannot handle this!')
log_error('** Exiting!')
sys.exit(1)
basenames.append(basename)
# check that files exist and there is sufficient output disk space.
check_valid_file_exists(args.input_filenames)
check_space(args.input_filenames, args.force)
if args.savegraph is not None:
graphsize = calculate_graphsize(args, 'countgraph')
check_space_for_graph(args.savegraph, graphsize, args.force)
# load or create counting table.
if args.loadgraph:
log_info('loading k-mer countgraph from {graph}', graph=args.loadgraph)
countgraph1 = Countgraph.load(args.loadgraph)
# load second counting table.
if args.loadgraph2:
log_info('loading k-mer countgraph from {graph}',
graph=args.loadgraph2)
countgraph2 = Countgraph.load(args.loadgraph2)
# make a list of all filenames and if they're paired or not;
# if we don't know if they're paired, default to allowing but not
# forcing pairing.
files = []
for element in filenames:
files.append([element, args.paired])
if args.unpaired_reads:
files.append([args.unpaired_reads, False])
#
# main loop: iterate over all files given, do diginorm.
#
for filename, require_paired in files:
if not args.single_output_file:
output_name = os.path.basename(filename) + '.keep'
outfp = open(output_name, 'wb')
outfp = get_file_writer(outfp, args.gzip, args.bzip)
screed_iter = clean_input_reads(screed.open(filename))
reader = broken_paired_reader(screed_iter,
min_length=args.ksize,
force_single=force_single,
require_paired=require_paired)
# actually do diginorm
for _, is_paired, read0, read1 in reader:
for record in snarf(is_paired, read0, read1, countgraph1,
countgraph2):
if record is not None:
write_record(record, outfp)
print("--- %s seconds ---" % (time.time() - start_time))
def main():
info('correct-reads.py', ['streaming'])
args = sanitize_help(get_parser()).parse_args()
###
if len(set(args.input_filenames)) != len(args.input_filenames):
print("Error: Cannot input the same filename multiple times.",
file=sys.stderr)
sys.exit(1)
###
report_on_config(args)
check_valid_file_exists(args.input_filenames)
check_space(args.input_filenames, args.force)
tablesize = calculate_graphsize(args, 'countgraph')
if args.savegraph:
check_space_for_graph(args.savegraph, tablesize,
args.force)
K = args.ksize
CUTOFF = args.cutoff
NORMALIZE_LIMIT = args.normalize_to
if args.loadgraph:
print('loading k-mer countgraph from', args.loadgraph, file=sys.stderr)
ct = Countgraph.load(args.loadgraph)
else:
print('making k-mer countgraph', file=sys.stderr)
ct = create_countgraph(args, multiplier=8 / (9. + 0.3))
tempdir = tempfile.mkdtemp('khmer', 'tmp', args.tempdir)
print('created temporary directory %s; use -T to change location'
% tempdir, file=sys.stderr)
aligner = khmer.ReadAligner(ct, args.cutoff, args.bits_theta)
# ### FIRST PASS ###
save_pass2_total = 0
n_bp = 0
n_reads = 0
written_bp = 0
written_reads = 0
corrected_reads = 0
pass2list = []
for filename in args.input_filenames:
pass2filename = os.path.basename(filename) + '.pass2'
pass2filename = os.path.join(tempdir, pass2filename)
if args.out is None:
corrfp = open(os.path.basename(filename) + '.corr', 'w')
else:
corrfp = args.out
pass2list.append((filename, pass2filename, corrfp))
screed_iter = screed.open(filename, parse_description=False)
pass2fp = open(pass2filename, 'w')
save_pass2 = 0
n = 0
paired_iter = broken_paired_reader(screed_iter, min_length=K,
force_single=args.ignore_pairs)
for n, is_pair, read1, read2 in paired_iter:
if n % 10000 == 0:
print('...', n, filename, save_pass2, n_reads, n_bp,
written_reads, written_bp, file=sys.stderr)
# we want to track paired reads here, to make sure that pairs
# are not split between first pass and second pass.
if is_pair:
n_reads += 2
n_bp += len(read1.sequence) + len(read2.sequence)
seq1 = read1.sequence.replace('N', 'A')
seq2 = read2.sequence.replace('N', 'A')
med1, _, _ = ct.get_median_count(seq1)
med2, _, _ = ct.get_median_count(seq2)
if med1 < NORMALIZE_LIMIT or med2 < NORMALIZE_LIMIT:
ct.consume(seq1)
ct.consume(seq2)
write_record_pair(read1, read2, pass2fp)
save_pass2 += 2
else:
is_aligned, new_seq1 = correct_sequence(aligner, seq1)
if is_aligned:
if new_seq1 != read1.sequence:
corrected_reads += 1
read1.sequence = new_seq1
if hasattr(read1, 'quality'):
fix_quality(read1)
is_aligned, new_seq2 = correct_sequence(aligner, seq2)
if is_aligned:
if new_seq2 != read2.sequence:
corrected_reads += 1
read2.sequence = new_seq2
if hasattr(read2, 'quality'):
fix_quality(read2)
write_record_pair(read1, read2, corrfp)
written_reads += 2
written_bp += len(read1)
written_bp += len(read2)
else:
n_reads += 1
n_bp += len(read1.sequence)
seq = read1.sequence.replace('N', 'A')
med, _, _ = ct.get_median_count(seq)
# has this portion of the graph saturated? if not,
# consume & save => pass2.
if med < NORMALIZE_LIMIT:
ct.consume(seq)
write_record(read1, pass2fp)
save_pass2 += 1
else: # trim!!
is_aligned, new_seq = correct_sequence(aligner, seq)
if is_aligned:
if new_seq != read1.sequence:
corrected_reads += 1
read1.sequence = new_seq
if hasattr(read1, 'quality'):
fix_quality(read1)
write_record(read1, corrfp)
written_reads += 1
written_bp += len(new_seq)
pass2fp.close()
print('%s: kept aside %d of %d from first pass, in %s'
% (filename, save_pass2, n, filename), file=sys.stderr)
save_pass2_total += save_pass2
# ### SECOND PASS. ###
skipped_n = 0
skipped_bp = 0
for _, pass2filename, corrfp in pass2list:
print(('second pass: looking at sequences kept aside in %s') %
pass2filename, file=sys.stderr)
# note that for this second pass, we don't care about paired
# reads - they will be output in the same order they're read in,
# so pairs will stay together if not orphaned. This is in contrast
# to the first loop.
for n, read in enumerate(screed.open(pass2filename,
parse_description=False)):
if n % 10000 == 0:
print('... x 2', n, pass2filename, written_reads,
written_bp, file=sys.stderr)
seq = read.sequence.replace('N', 'A')
med, _, _ = ct.get_median_count(seq)
# do we retain low-abundance components unchanged?
if med < NORMALIZE_LIMIT and args.variable_coverage:
write_record(read, corrfp)
written_reads += 1
written_bp += len(read.sequence)
skipped_n += 1
skipped_bp += len(read.sequence)
# otherwise, examine/correct.
else: # med >= NORMALIZE LIMIT or not args.variable_coverage
is_aligned, new_seq = correct_sequence(aligner, seq)
if is_aligned:
if new_seq != read.sequence:
corrected_reads += 1
read.sequence = new_seq
if hasattr(read, 'quality'):
fix_quality(read)
write_record(read, corrfp)
written_reads += 1
written_bp += len(new_seq)
print('removing %s' % pass2filename, file=sys.stderr)
os.unlink(pass2filename)
print('removing temp directory & contents (%s)' % tempdir, file=sys.stderr)
shutil.rmtree(tempdir)
n_passes = 1.0 + (float(save_pass2_total) / n_reads)
percent_reads_corrected = float(corrected_reads +
(n_reads - written_reads)) /\
n_reads * 100.0
print('read %d reads, %d bp' % (n_reads, n_bp,), file=sys.stderr)
print('wrote %d reads, %d bp' % (written_reads, written_bp,),
file=sys.stderr)
print('looked at %d reads twice (%.2f passes)' %
(save_pass2_total, n_passes), file=sys.stderr)
print('removed %d reads and corrected %d reads (%.2f%%)' %
(n_reads - written_reads, corrected_reads, percent_reads_corrected),
file=sys.stderr)
print('removed %.2f%% of bases (%d total)' %
((1 - (written_bp / float(n_bp))) * 100.0, n_bp - written_bp),
file=sys.stderr)
if args.variable_coverage:
percent_reads_hicov = 100.0 * float(n_reads - skipped_n) / n_reads
print('%d reads were high coverage (%.2f%%);' %
(n_reads - skipped_n, percent_reads_hicov), file=sys.stderr)
print(('skipped %d reads/%d bases because of low coverage')
% (skipped_n, skipped_bp), file=sys.stderr)
fp_rate = \
khmer.calc_expected_collisions(ct, args.force, max_false_pos=.8)
# for max_false_pos see Zhang et al., http://arxiv.org/abs/1309.2975
print('fp rate estimated to be {fpr:1.3f}'.format(fpr=fp_rate),
file=sys.stderr)
print('output in *.corr', file=sys.stderr)
if args.savegraph:
print("Saving k-mer countgraph to", args.savegraph, file=sys.stderr)
ct.save(args.savegraph)
