def bed(args):
"""
%prog bed fastafile kmer.dump.txt
Map kmers on FASTA.
"""
from jcvi.formats.fasta import rc, parse_fasta
p = OptionParser(bed.__doc__)
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
fastafile, dumpfile = args
fp = open(dumpfile)
KMERS = set()
for row in fp:
kmer = row.split()[0]
kmer_rc = rc(kmer)
KMERS.add(kmer)
KMERS.add(kmer_rc)
K = len(kmer)
logging.debug("Imported {} {}-mers".format(len(KMERS), K))
for name, seq in parse_fasta(fastafile):
name = name.split()[0]
for i in range(len(seq) - K):
if i % 5000000 == 0:
print("{}:{}".format(name, i), file=sys.stderr)
kmer = seq[i:i + K]
if kmer in KMERS:
print("\t".join(str(x) for x in (name, i, i + K, kmer)))
def bed(args):
"""
%prog bed fastafile kmer.dump.txt
Map kmers on FASTA.
"""
from jcvi.formats.fasta import rc, parse_fasta
p = OptionParser(bed.__doc__)
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
fastafile, dumpfile = args
fp = open(dumpfile)
KMERS = set()
for row in fp:
kmer = row.split()[0]
kmer_rc = rc(kmer)
KMERS.add(kmer)
KMERS.add(kmer_rc)
K = len(kmer)
logging.debug("Imported {} {}-mers".format(len(KMERS), K))
for name, seq in parse_fasta(fastafile):
name = name.split()[0]
for i in range(len(seq) - K):
if i % 5000000 == 0:
print >> sys.stderr, "{}:{}".format(name, i)
kmer = seq[i: i + K]
if kmer in KMERS:
print "\t".join(str(x) for x in (name, i, i + K, kmer))
def mid(args):
"""
%prog mid mappingfile
Produce a MID configuration file from primer mapping. The primer mapping can
be two or three columns, the third column is the optional 3`-primer.
"""
from jcvi.formats.fasta import rc
p = OptionParser(mid.__doc__)
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
mappingfile, = args
fp = open(mappingfile)
data = [row.split() for row in fp]
templatefile = "/usr/local/seq454-64_v2.6/config/MIDConfig.parse"
midfile = op.basename(templatefile)
fw = open(midfile, "w")
lines = open(templatefile).readlines()
insertline = 45
first, second = lines[:insertline], lines[insertline:]
for row in first:
fw.write(row)
# The inserted block
print >> fw, "MYMIDs\n{"
for atoms in data:
natoms = len(atoms)
assert natoms in (2, 3)
name, p5seq = atoms[:2]
line = ' mid = "{0}", "{1}", 1'.format(name, p5seq)
# Since most often I find p3 primers by parsing newbler progress file
# The string I grepped needs to be reverse-complemented
p3seq = atoms[2]
p3seq = rc(p3seq)
if natoms == 3:
line += ', "{0}"'.format(p3seq)
line += ';'
print >> fw, line
print >> fw, "}"
for row in second:
fw.write(row)
logging.debug("Barcodes written to `{0}`.".format(midfile))
def rc(self):
self.seq = rc(self.seq)
self.qual = self.qual[::-1]
def rc(self):
self.seq = rc(self.seq)
self.qual = self.qual[::-1]
