def main():
parser = build_construct_args()
parser.add_argument('-l',
'--lower-cutoff',
type=int,
dest='lower_cutoff',
default=DEFAULT_LOWER_CUTOFF)
parser.add_argument('-u',
'--upper-cutoff',
type=int,
dest='upper_cutoff',
default=DEFAULT_UPPER_CUTOFF)
parser.add_argument('output_filename')
parser.add_argument('input_filename')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print >>sys.stderr, "** WARNING: hashsize is default! " \
"You absodefly want to increase this!\n** " \
"Please read the docs!"
print >> sys.stderr, '\nPARAMETERS:'
print >> sys.stderr, ' - kmer size = %d \t\t(-k)' % args.ksize
print >> sys.stderr, ' - n hashes = %d \t\t(-N)' % args.n_hashes
print >>sys.stderr, ' - min hashsize = %-5.2g \t(-x)' % \
args.min_hashsize
print >> sys.stderr, ''
print >> sys.stderr, 'Estimated memory usage is %.2g bytes " \
"(n_hashes x min_hashsize)' % (args.n_hashes * args.min_hashsize)
print >> sys.stderr, '-' * 8
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
output = args.output_filename
input = args.input_filename
print 'lower cutoff:', args.lower_cutoff
print 'upper cutoff:', args.upper_cutoff
print 'Saving stoptags to %s' % output
print 'Loading sequences in %s' % input
###
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
ht.set_use_bigcount(True)
print 'consuming input', input
hb = ht.collect_high_abundance_kmers(input, args.lower_cutoff,
args.upper_cutoff)
print 'saving stoptags', output
hb.save_stop_tags(output)
def main():
parser = build_construct_args()
add_threading_args(parser)
parser.add_argument('datafile')
args = parser.parse_args()
report_on_config(args)
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
n_threads = int(args.n_threads)
config = khmer.get_config()
bufsz = config.get_reads_input_buffer_size()
config.set_reads_input_buffer_size(n_threads * 64 * 1024)
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT, n_threads)
filename = args.datafile
### first, load reads into hash table
rparser = khmer.ReadParser(filename, n_threads)
threads = []
print 'consuming input, round 1 --', filename
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=ht.consume_fasta_with_reads_parser,
args=(rparser, )
)
threads.append(t)
t.start()
for t in threads:
t.join()
fp_rate = khmer.calc_expected_collisions(ht)
print 'fp rate estimated to be %1.3f' % fp_rate
### now, count.
total = 0
total_unique = 0
for n, record in enumerate(screed.open(filename)):
total += 1
last_kmer = record.sequence[-K:]
count = ht.get(last_kmer)
if count == 1:
total_unique += 1
print 'singletons: %d unique; of %d total; %.3f' % \
(total_unique, total, total_unique/float(total))
def main():
parser = build_construct_args()
add_threading_args(parser)
parser.add_argument('datafile')
args = parser.parse_args()
report_on_config(args)
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
n_threads = int(args.n_threads)
config = khmer.get_config()
bufsz = config.get_reads_input_buffer_size()
config.set_reads_input_buffer_size(n_threads * 64 * 1024)
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT, n_threads)
filename = args.datafile
### first, load reads into hash table
rparser = khmer.ReadParser(filename, n_threads)
threads = []
print 'consuming input, round 1 --', filename
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=ht.consume_fasta_with_reads_parser,
args=(rparser, )
)
threads.append(t)
t.start()
for t in threads:
t.join()
fp_rate = khmer.calc_expected_collisions(ht)
print 'fp rate estimated to be %1.3f' % fp_rate
### now, count.
total = 0
total_unique = 0
for n, record in enumerate(screed.open(filename)):
total += 1
last_kmer = record.sequence[-K:]
count = ht.get(last_kmer)
if count == 1:
total_unique += 1
print 'singletons: %d unique; of %d total; %.3f' % \
(total_unique, total, total_unique/float(total))
def main():
parser = build_construct_args()
parser.add_argument('-l', '--lower-cutoff', type=int, dest='lower_cutoff',
default=DEFAULT_LOWER_CUTOFF)
parser.add_argument('-u', '--upper-cutoff', type=int, dest='upper_cutoff',
default=DEFAULT_UPPER_CUTOFF)
parser.add_argument('output_filename')
parser.add_argument('input_filename')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print >>sys.stderr, "** WARNING: hashsize is default! " \
"You absodefly want to increase this!\n** " \
"Please read the docs!"
print >>sys.stderr, '\nPARAMETERS:'
print >>sys.stderr, ' - kmer size = %d \t\t(-k)' % args.ksize
print >>sys.stderr, ' - n hashes = %d \t\t(-N)' % args.n_hashes
print >>sys.stderr, ' - min hashsize = %-5.2g \t(-x)' % \
args.min_hashsize
print >>sys.stderr, ''
print >>sys.stderr, 'Estimated memory usage is %.2g bytes " \
"(n_hashes x min_hashsize)' % (
args.n_hashes * args.min_hashsize)
print >>sys.stderr, '-' * 8
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
output = args.output_filename
input = args.input_filename
print 'lower cutoff:', args.lower_cutoff
print 'upper cutoff:', args.upper_cutoff
print 'Saving stoptags to %s' % output
print 'Loading sequences in %s' % input
###
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
ht.set_use_bigcount(True)
print 'consuming input', input
hb = ht.collect_high_abundance_kmers(input,
args.lower_cutoff,
args.upper_cutoff)
print 'saving stoptags', output
hb.save_stop_tags(output)
def main():
parser = build_construct_args()
parser.add_argument('-C', '--cutoff', type=int, dest='cutoff',
default=DEFAULT_DESIRED_COVERAGE)
parser.add_argument('-s', '--savehash', dest='savehash', default='')
parser.add_argument('-l', '--loadhash', dest='loadhash',
default='')
parser.add_argument('-R', '--report-to-file', dest='report_file',
type=argparse.FileType('w'))
parser.add_argument('input_filenames', nargs='+')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print>>sys.stderr, "** WARNING: hashsize is default! You absodefly want to increase this!\n** Please read the docs!"
print>>sys.stderr, '\nPARAMETERS:'
print>>sys.stderr, ' - kmer size = %d \t\t(-k)' % args.ksize
print>>sys.stderr, ' - n hashes = %d \t\t(-N)' % args.n_hashes
print>>sys.stderr, ' - min hashsize = %-5.2g \t(-x)' % args.min_hashsize
print>>sys.stderr, ''
print>>sys.stderr, 'Estimated memory usage is %.2g bytes (n_hashes x min_hashsize)' % (args.n_hashes * args.min_hashsize)
print>>sys.stderr, '-'*8
K=args.ksize
HT_SIZE=args.min_hashsize
N_HT=args.n_hashes
DESIRED_COVERAGE=args.cutoff
report_fp = args.report_file
filenames = args.input_filenames
if args.loadhash:
print 'loading hashtable from', args.loadhash
ht = khmer.load_counting_hash(args.loadhash)
else:
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
total = 0
discarded = 0
for input_filename in filenames:
output_name = os.path.basename(input_filename) + '.keep'
outfp = open(output_name, 'w')
n = -1
for n, record in enumerate(screed.open(input_filename)):
if n > 0 and n % 100000 == 0:
print '... kept', total - discarded, 'of', total, ', or', \
int(100. - discarded / float(total) * 100.), '%'
print '... in file', input_filename
if report_fp:
print>>report_fp, total, total - discarded, \
1. - (discarded / float(total))
report_fp.flush()
total += 1
if len(record.sequence) < K:
continue
seq = record.sequence.replace('N', 'A')
med, _, _ = ht.get_median_count(seq)
if med < DESIRED_COVERAGE:
ht.consume(seq)
outfp.write('>%s\n%s\n' % (record.name, record.sequence))
else:
discarded += 1
if -1 < n:
print 'DONE with', input_filename, '; kept', total - discarded, 'of',\
total, 'or', int(100. - discarded / float(total) * 100.), '%'
print 'output in', output_name
else: print 'SKIPPED empty file', input_filename
if args.savehash:
print 'Saving hashfile through', input_filename
print '...saving to', args.savehash
ht.save(args.savehash)
# 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' % fp_rate
if fp_rate > 0.20:
print >>sys.stderr, "**"
print >>sys.stderr, "** ERROR: the counting hash is too small for"
print >>sys.stderr, "** this data set. Increase hashsize/num ht."
print >>sys.stderr, "**"
print >>sys.stderr, "** Do not use these results!!"
sys.exit(-1)
def main():
parser = build_construct_args()
parser.add_argument('-C', '--cutoff', type=int, dest='cutoff',
default=DEFAULT_DESIRED_COVERAGE)
parser.add_argument('-p', '--paired', action='store_true')
parser.add_argument('-s', '--savehash', dest='savehash', default='')
parser.add_argument('-l', '--loadhash', dest='loadhash',
default='')
parser.add_argument('-R', '--report-to-file', dest='report_file',
type=argparse.FileType('w'))
parser.add_argument('input_filenames', nargs='+')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print>>sys.stderr, "** WARNING: hashsize is default! You absodefly want to increase this!\n** Please read the docs!"
print>>sys.stderr, '\nPARAMETERS:'
print>>sys.stderr, ' - kmer size = %d \t\t(-k)' % args.ksize
print>>sys.stderr, ' - n hashes = %d \t\t(-N)' % args.n_hashes
print>>sys.stderr, ' - min hashsize = %-5.2g \t(-x)' % args.min_hashsize
print>>sys.stderr, ' - paired = %s \t\t(-p)' % args.paired
print>>sys.stderr, ''
print>>sys.stderr, 'Estimated memory usage is %.2g bytes (n_hashes x min_hashsize)' % (args.n_hashes * args.min_hashsize)
print>>sys.stderr, '-'*8
K=args.ksize
HT_SIZE=args.min_hashsize
N_HT=args.n_hashes
DESIRED_COVERAGE=args.cutoff
report_fp = args.report_file
filenames = args.input_filenames
# In paired mode we read two records at a time
batch_size = 1
if args.paired:
batch_size = 2
if args.loadhash:
print 'loading hashtable from', args.loadhash
ht = khmer.load_counting_hash(args.loadhash)
else:
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
total = 0
discarded = 0
for input_filename in filenames:
output_name = os.path.basename(input_filename) + '.keep'
outfp = open(output_name, 'w')
n = -1
for n, batch in enumerate(batchwise(screed.open(input_filename), batch_size)):
if n > 0 and n % 100000 == 0:
print '... kept', total - discarded, 'of', total, ', or', \
int(100. - discarded / float(total) * 100.), '%'
print '... in file', input_filename
if report_fp:
print>>report_fp, total, total - discarded, \
1. - (discarded / float(total))
report_fp.flush()
total += batch_size
# If in paired mode, check that the reads are properly interleaved
if args.paired:
if not validpair(batch[0], batch[1]):
print >>sys.stderr, 'Error: Improperly interleaved pairs %s %s' % (batch[0].name, batch[1].name)
sys.exit(-1)
# Emit the batch of reads if any read passes the filter
# and all reads are longer than K
passed_filter = False
passed_length = True
for record in batch:
if len(record.sequence) < K:
passed_length = False
continue
seq = record.sequence.replace('N', 'A')
med, _, _ = ht.get_median_count(seq)
if med < DESIRED_COVERAGE:
ht.consume(seq)
passed_filter = True
# Emit records if any passed
if passed_length and passed_filter:
for record in batch:
if hasattr(record,'accuracy'):
outfp.write('@%s\n%s\n+\n%s\n' % (record.name,
record.sequence,
record.accuracy))
else:
outfp.write('>%s\n%s\n' % (record.name, record.sequence))
else:
discarded += batch_size
if -1 < n:
print 'DONE with', input_filename, '; kept', total - discarded, 'of',\
total, 'or', int(100. - discarded / float(total) * 100.), '%'
print 'output in', output_name
else: print 'SKIPPED empty file', input_filename
if args.savehash:
print 'Saving hashfile through', input_filename
print '...saving to', args.savehash
ht.save(args.savehash)
# 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' % fp_rate
if fp_rate > 0.20:
print >>sys.stderr, "**"
print >>sys.stderr, "** ERROR: the counting hash is too small for"
print >>sys.stderr, "** this data set. Increase hashsize/num ht."
print >>sys.stderr, "**"
print >>sys.stderr, "** Do not use these results!!"
sys.exit(-1)
def main():
parser = build_construct_args()
parser.add_argument('input_filenames', nargs='+')
parser.add_argument('read_filename')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print>>sys.stderr, "** WARNING: hashsize is default! You absodefly want to increase this!\n** Please read the docs!"
print>>sys.stderr, '\nPARAMETERS:'
print>>sys.stderr, ' - kmer size = %d \t\t(-k)' % args.ksize
print>>sys.stderr, ' - n hashes = %d \t\t(-N)' % args.n_hashes
print>>sys.stderr, ' - min hashsize = %-5.2g \t(-x)' % args.min_hashsize
print>>sys.stderr, ''
print>>sys.stderr, 'Estimated memory usage is %.2g bytes (n_hashes x min_hashsize)' % (args.n_hashes * args.min_hashsize)
print>>sys.stderr, '-'*8
K=args.ksize
HT_SIZE=args.min_hashsize
N_HT=args.n_hashes
inputlist = args.input_filenames
readsfile = args.read_filename
query_list = []
for n, inp_name in enumerate(inputlist):
# create a hashbits data structure
# ht = khmer.new_hashbits(K, HT_SIZE, N_HT)
# create a counting hash
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
outfile = os.path.basename(inp_name) + '.sweep3'
outfp = open(outfile, 'w')
query_list.append((ht, outfp))
for n, inp_name in enumerate(inputlist):
ht = query_list[n][0]
# load contigs, connect into N partitions
print 'loading input reads from', inp_name
ht.consume_fasta(inp_name)
# 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' % fp_rate
if fp_rate > 0.20:
print >>sys.stderr, "**"
print >>sys.stderr, "** ERROR: the counting hash is too small for"
print >>sys.stderr, "** this data set. Increase hashsize/num ht."
print >>sys.stderr, "**"
print >>sys.stderr, "** Do not use these results!!"
sys.exit(-1)
print 'starting sweep.'
n = 0
m = 0
for n, record in enumerate(screed.open(readsfile)):
if n % 10000 == 0:
print '...', n, m
for ht, outfp in query_list:
count = ht.get_median_count(record.sequence)[0]
if count:
outfp.write('>%s\n%s\n' % (record.name, record.sequence))
def main():
parser = build_construct_args(
"Output k-mer abundance distribution (single file version).")
add_threading_args(parser)
parser.add_argument('datafile')
parser.add_argument('histout')
parser.add_argument('-z', '--no-zero', dest='output_zero', default=True,
action='store_false',
help='Do not output 0-count bins')
parser.add_argument('-b', '--no-bigcount', dest='bigcount', default=True,
action='store_false',
help='Do not count k-mers past 255')
parser.add_argument('-s', '--squash', dest='squash_output', default=False,
action='store_true',
help='Overwrite output file if it exists')
parser.add_argument('--savehash', dest='savehash', default='')
args = parser.parse_args()
report_on_config(args)
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
n_threads = int(args.n_threads)
datafile = args.datafile
histout = args.histout
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT, n_threads)
ht.set_use_bigcount(args.bigcount)
print 'building tracking ht'
K = ht.ksize()
sizes = ht.hashsizes()
tracking = khmer._new_hashbits(K, sizes)
print 'K:', K
print 'HT sizes:', sizes
print 'outputting to', histout
config = khmer.get_config()
config.set_reads_input_buffer_size(n_threads * 64 * 1024)
# start loading
rparser = khmer.ReadParser(datafile, n_threads)
threads = []
print 'consuming input, round 1 --', datafile
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=ht.consume_fasta_with_reads_parser,
args=(rparser, )
)
threads.append(t)
t.start()
for t in threads:
t.join()
z_list = []
def do_abundance_dist(r):
z = ht.abundance_distribution_with_reads_parser(r, tracking)
z_list.append(z)
print 'preparing hist from %s...' % datafile
rparser = khmer.ReadParser(datafile, n_threads)
threads = []
print 'consuming input, round 2 --', datafile
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=do_abundance_dist,
args=(rparser,)
)
threads.append(t)
t.start()
for t in threads:
t.join()
assert len(z_list) == n_threads, len(z_list)
z = {}
for zz in z_list:
for i, count in enumerate(zz):
z[i] = z.get(i, 0) + count
total = sum(z.values())
if 0 == total:
print >>sys.stderr, \
"ERROR: abundance distribution is uniformly zero; " \
"nothing to report."
print >>sys.stderr, "\tPlease verify that the input files are valid."
sys.exit(-1)
fp = open(histout, 'w')
sofar = 0
for n, i in sorted(z.items()):
if i == 0 and not args.output_zero:
continue
sofar += i
frac = sofar / float(total)
print >>fp, n, i, sofar, round(frac, 3)
if sofar == total:
break
if args.savehash:
print 'Saving hashfile', args.savehash
print '...saving to', args.savehash
ht.save(args.savehash)
def main():
parser = build_construct_args()
parser.add_argument('-C',
'--cutoff',
type=int,
dest='cutoff',
default=DEFAULT_DESIRED_COVERAGE)
parser.add_argument('-p', '--paired', action='store_true')
parser.add_argument('-s', '--savehash', dest='savehash', default='')
parser.add_argument('-l', '--loadhash', dest='loadhash', default='')
parser.add_argument('-R',
'--report-to-file',
dest='report_file',
type=argparse.FileType('w'))
parser.add_argument('-f',
'--force-processing',
dest='force',
help='continue on next file if read errors are \
encountered',
action='store_true')
parser.add_argument('-d',
'--dump-frequency',
dest='dump_frequency',
type=int,
help='dump hashtable every d files',
default=-1)
parser.add_argument('input_filenames', nargs='+')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE and not args.loadhash:
print >>sys.stderr, \
"** WARNING: hashsize is default! " \
"You absodefly want to increase this!\n** " \
"Please read the docs!"
print >> sys.stderr, '\nPARAMETERS:'
print >>sys.stderr, \
' - kmer size = {ksize:d} \t\t(-k)'.format(ksize=args.ksize)
print >>sys.stderr, \
' - n hashes = {nhash:d} \t\t(-N)'.format(nhash=args.n_hashes)
print >>sys.stderr, \
' - min hashsize = {mh:-5.2g} \t(-x)'.format(mh=args.min_hashsize)
print >> sys.stderr, ' - paired = {pr} \t\t(-p)'.format(pr=args.paired)
print >> sys.stderr, ''
print >>sys.stderr, \
'Estimated memory usage is {prod:.2g} bytes \
(n_hashes x min_hashsize)' .format(prod=args.n_hashes
* args.min_hashsize)
print >> sys.stderr, '-' * 8
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
report_fp = args.report_file
filenames = args.input_filenames
force = args.force
dump_frequency = args.dump_frequency
# list to save error files along with throwing exceptions
if force is True:
corrupt_files = []
if args.loadhash:
print 'loading hashtable from', args.loadhash
ht = khmer.load_counting_hash(args.loadhash)
else:
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
total = 0
discarded = 0
for n, input_filename in enumerate(filenames):
output_name = os.path.basename(input_filename) + '.keep'
outfp = open(output_name, 'w')
total_acc = 0
discarded_acc = 0
try:
total_acc, discarded_acc = normalize_by_median(
input_filename, outfp, ht, args, report_fp)
except IOError as e:
handle_error(e, output_name, input_filename, ht)
if not force:
print >> sys.stderr, '** Exiting!'
sys.exit(-1)
else:
print >> sys.stderr, '*** Skipping error file, moving on...'
corrupt_files.append(input_filename)
pass
else:
if total_acc == 0 and discarded_acc == 0:
print 'SKIPPED empty file', input_filename
else:
total += total_acc
discarded += discarded_acc
print 'DONE with {inp}; kept {kept} of {total} or {perc:2}%'\
.format(inp=input_filename,
kept=total - discarded, total=total,
perc=int(100. - discarded / float(total) * 100.))
print 'output in', output_name
if dump_frequency > 0 and n > 0 and n % dump_frequency == 0:
print 'Backup: Saving hashfile through', input_filename
if args.savehash:
hashname = args.savehash
print '...saving to', hashname
else:
hashname = 'backup.ht'
print 'Nothing given for savehash, saving to', hashname
ht.save(hashname)
if args.savehash:
print 'Saving hashfile through', input_filename
print '...saving to', args.savehash
ht.save(args.savehash)
# 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 {fpr:1.3f}'.format(fpr=fp_rate)
if force and len(corrupt_files) > 0:
print >> sys.stderr, "** WARNING: Finished with errors!"
print >> sys.stderr, "** IOErrors occurred in the following files:"
print >> sys.stderr, "\t", " ".join(corrupt_files)
if fp_rate > 0.20:
print >> sys.stderr, "**"
print >> sys.stderr, "** ERROR: the counting hash is too small for"
print >> sys.stderr, "** this data set. Increase hashsize/num ht."
print >> sys.stderr, "**"
print >> sys.stderr, "** Do not use these results!!"
sys.exit(-1)
def main():
parser = build_construct_args()
parser.add_argument("-C", "--cutoff", type=int, dest="cutoff", default=DEFAULT_MINIMUM_COVERAGE)
parser.add_argument("-s", "--savehash", dest="savehash", default="")
parser.add_argument("-l", "--loadhash", dest="loadhash", default="")
parser.add_argument("input_filenames", nargs="+")
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE and not args.loadhash:
print >> sys.stderr, "** WARNING: hashsize is default! " "You absodefly want to increase this!\n** " "Please read the docs!"
print >> sys.stderr, "\nPARAMETERS:"
print >> sys.stderr, " - kmer size = %d \t\t(-k)" % args.ksize
print >> sys.stderr, " - n hashes = %d \t\t(-N)" % args.n_hashes
print >> sys.stderr, " - min hashsize = %-5.2g \t(-x)" % args.min_hashsize
print >> sys.stderr, ""
print >> sys.stderr, "Estimated memory usage is %.2g bytes " "(n_hashes x min_hashsize)" % (
args.n_hashes * args.min_hashsize
)
print >> sys.stderr, "-" * 8
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
DESIRED_COVERAGE = args.cutoff
filenames = args.input_filenames
if args.loadhash:
print "loading hashtable from", args.loadhash
ht = khmer.load_counting_hash(args.loadhash)
else:
print "making hashtable"
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
total = 0
discarded = 0
for input_filename in filenames:
output_name = os.path.basename(input_filename) + ".minkeep"
outfp = open(output_name, "w")
for n, record in enumerate(screed.open(input_filename)):
if n > 0 and n % 10000 == 0:
print "... kept", total - discarded, "of", total, ", or", int(
100.0 - discarded / float(total) * 100.0
), "%"
print "... in file", input_filename
total += 1
if len(record.sequence) < K:
continue
seq = record.sequence.replace("N", "A")
mincount = ht.get_min_count(seq)
if mincount < DESIRED_COVERAGE:
ht.consume(seq)
outfp.write(">%s\n%s\n" % (record.name, record.sequence))
else:
discarded += 1
print "DONE with", input_filename, "; kept", total - discarded, "of", total, "or", int(
100.0 - discarded / float(total) * 100.0
), "%"
print "output in", output_name
if args.savehash:
print "Saving hashfile through", input_filename
print "...saving to", args.savehash
ht.save(args.savehash)
# 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" % fp_rate
if fp_rate > 0.20:
print >> sys.stderr, "**"
print >> sys.stderr, "** ERROR: the counting hash is too small for"
print >> sys.stderr, "** this data set. Increase hashsize/num ht."
print >> sys.stderr, "**"
print >> sys.stderr, "** Do not use these results!!"
sys.exit(-1)
def main():
parser = build_construct_args()
parser.add_argument('-C', '--cutoff', type=int, dest='cutoff',
default=DEFAULT_DESIRED_COVERAGE)
parser.add_argument('-p', '--paired', action='store_true')
parser.add_argument('-s', '--savehash', dest='savehash', default='')
parser.add_argument('-l', '--loadhash', dest='loadhash',
default='')
parser.add_argument('-R', '--report-to-file', dest='report_file',
type=argparse.FileType('w'))
parser.add_argument('input_filenames', nargs='+')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print >>sys.stderr, \
"** WARNING: hashsize is default! " \
"You absodefly want to increase this!\n** " \
"Please read the docs!"
print >>sys.stderr, '\nPARAMETERS:'
print >>sys.stderr, ' - kmer size = %d \t\t(-k)' % args.ksize
print >>sys.stderr, ' - n hashes = %d \t\t(-N)' % args.n_hashes
print >>sys.stderr, \
' - min hashsize = %-5.2g \t(-x)' % args.min_hashsize
print >>sys.stderr, ' - paired = %s \t\t(-p)' % args.paired
print >>sys.stderr, ''
print >>sys.stderr, \
'Estimated memory usage is %.2g bytes (n_hashes x min_hashsize)' \
% (args.n_hashes * args.min_hashsize)
print >>sys.stderr, '-' * 8
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
DESIRED_COVERAGE = args.cutoff
report_fp = args.report_file
filenames = args.input_filenames
# In paired mode we read two records at a time
batch_size = 1
if args.paired:
batch_size = 2
if args.loadhash:
print 'loading hashtable from', args.loadhash
ht = khmer.load_counting_hash(args.loadhash)
else:
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
total = 0
discarded = 0
for input_filename in filenames:
output_name = os.path.basename(input_filename) + '.keep'
outfp = open(output_name, 'w')
n = -1
for n, batch in enumerate(batchwise(screed.open(
input_filename), batch_size)):
if n > 0 and n % 10000 == 0:
print '... kept', total - discarded, 'of', total, ', or', \
int(100. - discarded / float(total) * 100.), '%'
print '... in file', input_filename
if report_fp:
print>>report_fp, total, total - discarded, \
1. - (discarded / float(total))
report_fp.flush()
total += batch_size
# If in paired mode, check that the reads are properly interleaved
if args.paired:
if not validpair(batch[0], batch[1]):
print >>sys.stderr, \
'Error: Improperly interleaved pairs %s %s' \
% (batch[0].name, batch[1].name)
sys.exit(-1)
# Emit the batch of reads if any read passes the filter
# and all reads are longer than K
passed_filter = False
passed_length = True
for record in batch:
if len(record.sequence) < K:
passed_length = False
continue
seq = record.sequence.replace('N', 'A')
med, _, _ = ht.get_median_count(seq)
if med < DESIRED_COVERAGE:
ht.consume(seq)
passed_filter = True
# Emit records if any passed
if passed_length and passed_filter:
for record in batch:
if hasattr(record, 'accuracy'):
outfp.write('@%s\n%s\n+\n%s\n' % (record.name,
record.sequence,
record.accuracy))
else:
outfp.write(
'>%s\n%s\n' % (record.name, record.sequence))
else:
discarded += batch_size
if -1 < n:
print \
'DONE with', input_filename, '; kept', total - discarded, \
'of', total, 'or', \
int(100. - discarded / float(total) * 100.), '%'
print 'output in', output_name
else:
print 'SKIPPED empty file', input_filename
if args.savehash:
print 'Saving hashfile through', input_filename
print '...saving to', args.savehash
ht.save(args.savehash)
# 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' % fp_rate
if fp_rate > 0.20:
print >>sys.stderr, "**"
print >>sys.stderr, "** ERROR: the counting hash is too small for"
print >>sys.stderr, "** this data set. Increase hashsize/num ht."
print >>sys.stderr, "**"
print >>sys.stderr, "** Do not use these results!!"
sys.exit(-1)
def main():
parser = build_construct_args()
parser.add_argument('-C',
'--cutoff',
type=int,
dest='cutoff',
default=DEFAULT_DESIRED_COVERAGE)
parser.add_argument('-s', '--savehash', dest='savehash', default='')
parser.add_argument('-l', '--loadhash', dest='loadhash', default='')
parser.add_argument('input_filenames', nargs='+')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print >> sys.stderr, "** WARNING: hashsize is default! You absodefly want to increase this!\n** Please read the docs!"
print >> sys.stderr, '\nPARAMETERS:'
print >> sys.stderr, ' - kmer size = %d \t\t(-k)' % args.ksize
print >> sys.stderr, ' - n hashes = %d \t\t(-N)' % args.n_hashes
print >> sys.stderr, ' - min hashsize = %-5.2g \t(-x)' % args.min_hashsize
print >> sys.stderr, ''
print >> sys.stderr, 'Estimated memory usage is %.2g bytes (n_hashes x min_hashsize)' % (
args.n_hashes * args.min_hashsize)
print >> sys.stderr, '-' * 8
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
DESIRED_COVERAGE = args.cutoff
filenames = args.input_filenames
if args.loadhash:
print 'loading hashtable from', args.loadhash
ht = khmer.load_counting_hash(args.loadhash)
else:
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
total = 0
discarded = 0
for input_filename in filenames:
output_name = os.path.basename(input_filename) + '.keepkad'
outfp = open(output_name, 'w')
for n, record in enumerate(screed.open(input_filename)):
if n > 0 and n % 10000 == 0:
print '... kept', total - discarded, 'of', total, ', or', \
int(100. - discarded / float(total) * 100.), '%'
print '... in file', input_filename
total += 1
if len(record.sequence) < K:
continue
seq = record.sequence.replace('N', 'A')
kad = ht.get_kadian_count(seq)
if kad < DESIRED_COVERAGE:
ht.consume(seq)
outfp.write('>%s\n%s\n' % (record.name, record.sequence))
else:
discarded += 1
print 'DONE with', input_filename, '; kept', total - discarded, 'of',\
total, 'or', int(100. - discarded / float(total) * 100.), '%'
print 'output in', output_name
if args.savehash:
print 'Saving hashfile through', input_filename
print '...saving to', args.savehash
ht.save(args.savehash)
# 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' % fp_rate
if fp_rate > 0.20:
print >> sys.stderr, "**"
print >> sys.stderr, "** ERROR: the counting hash is too small for"
print >> sys.stderr, "** this data set. Increase hashsize/num ht."
print >> sys.stderr, "**"
print >> sys.stderr, "** Do not use these results!!"
sys.exit(-1)
def main():
parser = build_construct_args()
parser.add_argument("output_filename")
parser.add_argument("input_filenames", nargs="+")
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print >> sys.stderr, "** WARNING: hashsize is default! You absodefly want to increase this!\n** Please read the docs!"
print >> sys.stderr, "\nPARAMETERS:"
print >> sys.stderr, " - kmer size = %d \t\t(-k)" % args.ksize
print >> sys.stderr, " - n hashes = %d \t\t(-N)" % args.n_hashes
print >> sys.stderr, " - min hashsize = %-5.2g \t(-x)" % args.min_hashsize
print >> sys.stderr, ""
print >> sys.stderr, "Estimated memory usage is %.2g bytes (n_hashes x min_hashsize)" % (
args.n_hashes * args.min_hashsize
)
print >> sys.stderr, "-" * 8
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
base = args.output_filename
filenames = args.input_filenames
print "Saving hashtable to %s" % base
print "Loading kmers from sequences in %s" % repr(filenames)
###
print "making hashtable"
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
ht.set_use_bigcount(True)
for n, filename in enumerate(filenames):
print "consuming input", filename
ht.consume_fasta(filename)
if n > 0 and n % 10 == 0:
print "mid-save", base
ht.save(base)
open(base + ".info", "w").write("through %s" % filename)
print "saving", base
ht.save(base)
info_fp = open(base + ".info", "w")
info_fp.write("through end: %s\n" % filename)
# 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" % fp_rate
print >> info_fp, "fp rate estimated to be %1.3f" % fp_rate
if fp_rate > 0.20:
print >> sys.stderr, "**"
print >> sys.stderr, "** ERROR: the counting hash is too small for"
print >> sys.stderr, "** this data set. Increase hashsize/num ht."
print >> sys.stderr, "**"
sys.exit(-1)
def main():
parser = build_construct_args(
"Output k-mer abundance distribution (single file version).")
add_threading_args(parser)
parser.add_argument('datafile')
parser.add_argument('histout')
parser.add_argument('-z',
'--no-zero',
dest='output_zero',
default=True,
action='store_false',
help='Do not output 0-count bins')
parser.add_argument('-b',
'--no-bigcount',
dest='bigcount',
default=True,
action='store_false',
help='Do not count k-mers past 255')
parser.add_argument('-s',
'--squash',
dest='squash_output',
default=False,
action='store_true',
help='Overwrite output file if it exists')
parser.add_argument('--savehash', dest='savehash', default='')
args = parser.parse_args()
report_on_config(args)
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
n_threads = int(args.n_threads)
datafile = args.datafile
histout = args.histout
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT, n_threads)
ht.set_use_bigcount(args.bigcount)
print 'building tracking ht'
K = ht.ksize()
sizes = ht.hashsizes()
tracking = khmer._new_hashbits(K, sizes)
print 'K:', K
print 'HT sizes:', sizes
print 'outputting to', histout
config = khmer.get_config()
config.set_reads_input_buffer_size(n_threads * 64 * 1024)
# start loading
rparser = khmer.ReadParser(datafile, n_threads)
threads = []
print 'consuming input, round 1 --', datafile
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=ht.consume_fasta_with_reads_parser,
args=(rparser, )
)
threads.append(t)
t.start()
for t in threads:
t.join()
z_list = []
def do_abundance_dist(r):
z = ht.abundance_distribution_with_reads_parser(r, tracking)
z_list.append(z)
print 'preparing hist from %s...' % datafile
rparser = khmer.ReadParser(datafile, n_threads)
threads = []
print 'consuming input, round 2 --', datafile
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=do_abundance_dist,
args=(rparser,)
)
threads.append(t)
t.start()
for t in threads:
t.join()
assert len(z_list) == n_threads, len(z_list)
z = {}
for zz in z_list:
for i, count in enumerate(zz):
z[i] = z.get(i, 0) + count
total = sum(z.values())
if 0 == total:
print >>sys.stderr, \
"ERROR: abundance distribution is uniformly zero; " \
"nothing to report."
print >> sys.stderr, "\tPlease verify that the input files are valid."
sys.exit(-1)
fp = open(histout, 'w')
sofar = 0
for n, i in sorted(z.items()):
if i == 0 and not args.output_zero:
continue
sofar += i
frac = sofar / float(total)
print >> fp, n, i, sofar, round(frac, 3)
if sofar == total:
break
if args.savehash:
print 'Saving hashfile', args.savehash
print '...saving to', args.savehash
ht.save(args.savehash)
def main():
parser = build_construct_args()
parser.add_argument('output_filename')
parser.add_argument('input_filenames', nargs='+')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print>>sys.stderr, "** WARNING: hashsize is default! You absodefly want to increase this!\n** Please read the docs!"
print>>sys.stderr, '\nPARAMETERS:'
print>>sys.stderr, ' - kmer size = %d \t\t(-k)' % args.ksize
print>>sys.stderr, ' - n hashes = %d \t\t(-N)' % args.n_hashes
print>>sys.stderr, ' - min hashsize = %-5.2g \t(-x)' % args.min_hashsize
print>>sys.stderr, ''
print>>sys.stderr, 'Estimated memory usage is %.2g bytes (n_hashes x min_hashsize)' % (args.n_hashes * args.min_hashsize)
print>>sys.stderr, '-'*8
K=args.ksize
HT_SIZE=args.min_hashsize
N_HT=args.n_hashes
base = args.output_filename
filenames = args.input_filenames
print 'Saving hashtable to %s' % base
print 'Loading kmers from sequences in %s' % repr(filenames)
###
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
ht.set_use_bigcount(True)
for n, filename in enumerate(filenames):
print 'consuming input', filename
ht.consume_fasta(filename)
if n > 0 and n % 10 == 0:
print 'mid-save', base
ht.save(base)
open(base + '.info', 'w').write('through %s' % filename)
print 'saving', base
ht.save(base)
info_fp = open(base + '.info', 'w')
info_fp.write('through end: %s\n' % filename)
# 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' % fp_rate
print >>info_fp, 'fp rate estimated to be %1.3f' % fp_rate
if fp_rate > 0.20:
print >>sys.stderr, "**"
print >>sys.stderr, "** ERROR: the counting hash is too small for"
print >>sys.stderr, "** this data set. Increase hashsize/num ht."
print >>sys.stderr, "**"
sys.exit(-1)
print 'DONE.'
def main():
parser = build_construct_args()
add_threading_args(parser)
parser.add_argument('output_filename')
parser.add_argument('input_filenames', nargs='+')
parser.add_argument('-b',
'--no-bigcount',
dest='bigcount',
default=True,
action='store_false',
help='Do not count k-mers past 255')
args = parser.parse_args()
report_on_config(args)
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
base = args.output_filename
filenames = args.input_filenames
n_threads = int(args.n_threads)
print 'Saving hashtable to %s' % base
print 'Loading kmers from sequences in %s' % repr(filenames)
#
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT, n_threads)
ht.set_use_bigcount(args.bigcount)
config = khmer.get_config()
bufsz = config.get_reads_input_buffer_size()
config.set_reads_input_buffer_size(n_threads * 64 * 1024)
for n, filename in enumerate(filenames):
rparser = khmer.ReadParser(filename, n_threads)
threads = []
print 'consuming input', filename
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=ht.consume_fasta_with_reads_parser,
args=(rparser, )
)
threads.append(t)
t.start()
for t in threads:
t.join()
if n > 0 and n % 10 == 0:
print 'mid-save', base
ht.save(base)
open(base + '.info', 'w').write('through %s' % filename)
print 'saving', base
ht.save(base)
info_fp = open(base + '.info', 'w')
info_fp.write('through end: %s\n' % filename)
# 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' % fp_rate
print >> info_fp, 'fp rate estimated to be %1.3f' % fp_rate
if fp_rate > 0.20:
print >> sys.stderr, "**"
print >> sys.stderr, "** ERROR: the counting hash is too small for"
print >> sys.stderr, "** this data set. Increase hashsize/num ht."
print >> sys.stderr, "**"
sys.exit(-1)
print 'DONE.'
def main():
parser = build_construct_args(
"Filter k-mers at the given abundance (inmem version).")
add_threading_args(parser)
parser.add_argument('--cutoff',
'-C',
dest='cutoff',
default=DEFAULT_CUTOFF,
type=int,
help="Trim at k-mers below this abundance.")
parser.add_argument('--savehash', dest='savehash', default='')
parser.add_argument('datafile')
args = parser.parse_args()
report_on_config(args)
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
n_threads = int(args.n_threads)
config = khmer.get_config()
bufsz = config.get_reads_input_buffer_size()
config.set_reads_input_buffer_size(n_threads * 64 * 1024)
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT, n_threads)
filename = args.datafile
# first, load reads into hash table
rparser = khmer.ReadParser(filename, n_threads)
threads = []
print 'consuming input, round 1 --', filename
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=ht.consume_fasta_with_reads_parser,
args=(rparser, )
)
threads.append(t)
t.start()
for t in threads:
t.join()
fp_rate = khmer.calc_expected_collisions(ht)
print 'fp rate estimated to be %1.3f' % fp_rate
# now, trim.
# the filtering function.
def process_fn(record):
name = record['name']
seq = record['sequence']
if 'N' in seq:
return None, None
trim_seq, trim_at = ht.trim_on_abundance(seq, args.cutoff)
if trim_at >= K:
return name, trim_seq
return None, None
# the filtering loop
print 'filtering', filename
outfile = os.path.basename(filename) + '.abundfilt'
outfp = open(outfile, 'w')
tsp = ThreadedSequenceProcessor(process_fn)
tsp.start(verbose_loader(filename), outfp)
print 'output in', outfile
if args.savehash:
print 'Saving hashfile', args.savehash
print '...saving to', args.savehash
ht.save(args.savehash)
def main():
parser = build_construct_args()
add_threading_args(parser)
parser.add_argument('output_filename')
parser.add_argument('input_filenames', nargs='+')
parser.add_argument('-b', '--no-bigcount', dest='bigcount', default=True,
action='store_false',
help='Do not count k-mers past 255')
args = parser.parse_args()
report_on_config(args)
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
base = args.output_filename
filenames = args.input_filenames
n_threads = int(args.n_threads)
print 'Saving hashtable to %s' % base
print 'Loading kmers from sequences in %s' % repr(filenames)
###
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT, n_threads)
ht.set_use_bigcount(args.bigcount)
config = khmer.get_config()
bufsz = config.get_reads_input_buffer_size()
config.set_reads_input_buffer_size(n_threads * 64 * 1024)
for n, filename in enumerate(filenames):
rparser = khmer.ReadParser(filename, n_threads)
threads = []
print 'consuming input', filename
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=ht.consume_fasta_with_reads_parser,
args=(rparser, )
)
threads.append(t)
t.start()
for t in threads:
t.join()
if n > 0 and n % 10 == 0:
print 'mid-save', base
ht.save(base)
open(base + '.info', 'w').write('through %s' % filename)
print 'saving', base
ht.save(base)
info_fp = open(base + '.info', 'w')
info_fp.write('through end: %s\n' % filename)
# 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' % fp_rate
print >>info_fp, 'fp rate estimated to be %1.3f' % fp_rate
if fp_rate > 0.20:
print >>sys.stderr, "**"
print >>sys.stderr, "** ERROR: the counting hash is too small for"
print >>sys.stderr, "** this data set. Increase hashsize/num ht."
print >>sys.stderr, "**"
sys.exit(-1)
print 'DONE.'
def main():
parser = build_construct_args()
parser.add_argument('-C', '--cutoff', type=int, dest='cutoff',
default=DEFAULT_DESIRED_COVERAGE)
parser.add_argument('-p', '--paired', action='store_true')
parser.add_argument('-s', '--savehash', dest='savehash', default='')
parser.add_argument('-l', '--loadhash', dest='loadhash',
default='')
parser.add_argument('-R', '--report-to-file', dest='report_file',
type=argparse.FileType('w'))
parser.add_argument('-f', '--force-processing', dest='force',
help='continue on next file if read errors are \
encountered', action='store_true')
parser.add_argument('-d', '--dump-frequency', dest='dump_frequency',
type=int, help='dump hashtable every d files',
default=-1)
parser.add_argument('input_filenames', nargs='+')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print >>sys.stderr, \
"** WARNING: hashsize is default! " \
"You absodefly want to increase this!\n** " \
"Please read the docs!"
print >>sys.stderr, '\nPARAMETERS:'
print >>sys.stderr, \
' - kmer size = {ksize:d} \t\t(-k)'.format(ksize=args.ksize)
print >>sys.stderr, \
' - n hashes = {nhash:d} \t\t(-N)'.format(nhash=args.n_hashes)
print >>sys.stderr, \
' - min hashsize = {mh:-5.2g} \t(-x)'.format(mh=args.min_hashsize)
print >>sys.stderr, ' - paired = {pr} \t\t(-p)'.format(pr=args.paired)
print >>sys.stderr, ''
print >>sys.stderr, \
'Estimated memory usage is {prod:.2g} bytes \
(n_hashes x min_hashsize)'.format(prod=args.n_hashes*args.min_hashsize)
print >>sys.stderr, '-' * 8
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
report_fp = args.report_file
filenames = args.input_filenames
force=args.force
dump_frequency = args.dump_frequency
# list to save error files along with throwing exceptions
if force == True:
corrupt_files = []
if args.loadhash:
print 'loading hashtable from', args.loadhash
ht = khmer.load_counting_hash(args.loadhash)
else:
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
total = 0
discarded = 0
for n, input_filename in enumerate(filenames):
output_name = os.path.basename(input_filename) + '.keep'
outfp = open(output_name, 'w')
total_acc = 0
discarded_acc = 0
try:
total_acc, discarded_acc = normalize_by_median(input_filename,
outfp, ht, args,
report_fp)
except IOError as e:
handle_error(e, output_name, input_filename, ht)
if not force:
print >>sys.stderr, '** Exiting!'
sys.exit(-1)
else:
print >>sys.stderr, '*** Skipping error file, moving on...'
corrupt_files.append(input_filename)
pass
else:
if total_acc == 0 and discarded_acc == 0:
print 'SKIPPED empty file', input_filename
else:
total += total_acc
discarded += discarded_acc
print 'DONE with {inp}; kept {kept} of {total} or {perc:2}%'\
.format(inp=input_filename,
kept=total-discarded, total=total,
perc=int(100. - discarded / float(total) * 100.))
print 'output in', output_name
if dump_frequency > 0 and n > 0 and n % dump_frequency == 0:
print 'Backup: Saving hashfile through', input_filename
if args.savehash:
hashname = args.savehash
print '...saving to', hashname
else:
hashname = 'backup.ht'
print 'Nothing given for savehash, saving to', hashname
ht.save(hashname)
if args.savehash:
print 'Saving hashfile through', input_filename
print '...saving to', args.savehash
ht.save(args.savehash)
# 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 {fpr:1.3f}'.format(fpr=fp_rate)
if force and len(corrupt_files) > 0:
print >>sys.stderr, "** WARNING: Finished with errors!"
print >>sys.stderr, "** IOErrors occurred in the following files:"
print >>sys.stderr, "\t", " ".join(corrupt_files)
if fp_rate > 0.20:
print >>sys.stderr, "**"
print >>sys.stderr, "** ERROR: the counting hash is too small for"
print >>sys.stderr, "** this data set. Increase hashsize/num ht."
print >>sys.stderr, "**"
print >>sys.stderr, "** Do not use these results!!"
sys.exit(-1)
def main():
parser = build_construct_args(
"Filter k-mers at the given abundance (inmem version).")
add_threading_args(parser)
parser.add_argument('--cutoff', '-C', dest='cutoff',
default=DEFAULT_CUTOFF, type=int,
help="Trim at k-mers below this abundance.")
parser.add_argument('--savehash', dest='savehash', default='')
parser.add_argument('datafile')
args = parser.parse_args()
report_on_config(args)
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
n_threads = int(args.n_threads)
config = khmer.get_config()
bufsz = config.get_reads_input_buffer_size()
config.set_reads_input_buffer_size(n_threads * 64 * 1024)
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT, n_threads)
filename = args.datafile
### first, load reads into hash table
rparser = khmer.ReadParser(filename, n_threads)
threads = []
print 'consuming input, round 1 --', filename
for tnum in xrange(n_threads):
t = \
threading.Thread(
target=ht.consume_fasta_with_reads_parser,
args=(rparser, )
)
threads.append(t)
t.start()
for t in threads:
t.join()
fp_rate = khmer.calc_expected_collisions(ht)
print 'fp rate estimated to be %1.3f' % fp_rate
### now, trim.
### the filtering function.
def process_fn(record):
name = record['name']
seq = record['sequence']
if 'N' in seq:
return None, None
trim_seq, trim_at = ht.trim_on_abundance(seq, args.cutoff)
if trim_at >= K:
return name, trim_seq
return None, None
### the filtering loop
print 'filtering', filename
outfile = os.path.basename(filename) + '.abundfilt'
outfp = open(outfile, 'w')
tsp = ThreadedSequenceProcessor(process_fn)
tsp.start(verbose_loader(filename), outfp)
print 'output in', outfile
if args.savehash:
print 'Saving hashfile', args.savehash
print '...saving to', args.savehash
ht.save(args.savehash)
def main():
parser = build_construct_args()
parser.add_argument('output_filename')
parser.add_argument('input_filenames', nargs='+')
args = parser.parse_args()
if not args.quiet:
if args.min_hashsize == DEFAULT_MIN_HASHSIZE:
print >> sys.stderr, "** WARNING: hashsize is default! You absodefly want to increase this!\n** Please read the docs!"
print >> sys.stderr, '\nPARAMETERS:'
print >> sys.stderr, ' - kmer size = %d \t\t(-k)' % args.ksize
print >> sys.stderr, ' - n hashes = %d \t\t(-N)' % args.n_hashes
print >> sys.stderr, ' - min hashsize = %-5.2g \t(-x)' % args.min_hashsize
print >> sys.stderr, ''
print >> sys.stderr, 'Estimated memory usage is %.2g bytes (n_hashes x min_hashsize)' % (
args.n_hashes * args.min_hashsize)
print >> sys.stderr, '-' * 8
K = args.ksize
HT_SIZE = args.min_hashsize
N_HT = args.n_hashes
base = args.output_filename
filenames = args.input_filenames
print 'Saving hashtable to %s' % base
print 'Loading kmers from sequences in %s' % repr(filenames)
###
print 'making hashtable'
ht = khmer.new_counting_hash(K, HT_SIZE, N_HT)
ht.set_use_bigcount(True)
for n, filename in enumerate(filenames):
print 'consuming input', filename
ht.consume_fasta(filename)
if n > 0 and n % 10 == 0:
print 'mid-save', base
ht.save(base)
open(base + '.info', 'w').write('through %s' % filename)
print 'saving', base
ht.save(base)
info_fp = open(base + '.info', 'w')
info_fp.write('through end: %s\n' % filename)
# 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' % fp_rate
print >> info_fp, 'fp rate estimated to be %1.3f' % fp_rate
if fp_rate > 0.20:
print >> sys.stderr, "**"
print >> sys.stderr, "** ERROR: the counting hash is too small for"
print >> sys.stderr, "** this data set. Increase hashsize/num ht."
print >> sys.stderr, "**"
sys.exit(-1)
print 'DONE.'
