def bisect(args):
"""
%prog bisect acc accession.fasta
determine the version of the accession by querying entrez, based on a fasta file.
This proceeds by a sequential search from xxxx.1 to the latest record.
"""
p = OptionParser(bisect.__doc__)
p.set_email()
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
acc, fastafile = args
arec = get_first_rec(fastafile)
valid = None
for i in range(1, 100):
term = "%s.%d" % (acc, i)
try:
query = list(batch_entrez([term], email=opts.email))
except AssertionError as e:
logging.debug("no records found for %s. terminating." % term)
return
id, term, handle = query[0]
brec = next(SeqIO.parse(handle, "fasta"))
match = print_first_difference(arec,
brec,
ignore_case=True,
ignore_N=True,
rc=True)
if match:
valid = term
break
if valid:
printf()
printf("[green]{} matches the sequence in `{}`".format(
valid, fastafile))
def bisect(args):
"""
%prog bisect acc accession.fasta
determine the version of the accession by querying entrez, based on a fasta file.
This proceeds by a sequential search from xxxx.1 to the latest record.
"""
p = OptionParser(bisect.__doc__)
p.set_email()
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
acc, fastafile = args
arec = get_first_rec(fastafile)
valid = None
for i in range(1, 100):
term = "%s.%d" % (acc, i)
try:
query = list(batch_entrez([term], email=opts.email))
except AssertionError as e:
logging.debug("no records found for %s. terminating." % term)
return
id, term, handle = query[0]
brec = SeqIO.parse(handle, "fasta").next()
match = print_first_difference(arec, brec, ignore_case=True,
ignore_N=True, rc=True)
if match:
valid = term
break
if valid:
print
print green("%s matches the sequence in `%s`" % (valid, fastafile))
