def longest(args):
"""
%prog longest bedfile fastafile
Select longest feature within overlapping piles.
"""
from jcvi.formats.sizes import Sizes
p = OptionParser(longest.__doc__)
p.add_option("--maxsize", default=20000, type="int", help="Limit max size")
p.add_option("--minsize", default=60, type="int", help="Limit min size")
p.add_option("--precedence",
default="Medtr",
help="Accessions with prefix take precedence")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
bedfile, fastafile = args
maxsize = opts.maxsize
minsize = opts.minsize
prec = opts.precedence
mergedbed = mergeBed(bedfile, nms=True)
sizes = Sizes(fastafile).mapping
bed = Bed(mergedbed)
pf = bedfile.rsplit(".", 1)[0]
ids = set()
for b in bed:
accns = b.accn.split(";")
prec_accns = [x for x in accns if x.startswith(prec)]
if prec_accns:
accns = prec_accns
accn_sizes = [(sizes.get(x, 0), x) for x in accns]
accn_sizes = [(size, x) for size, x in accn_sizes if size < maxsize]
if not accn_sizes:
continue
max_size, max_accn = max(accn_sizes)
if max_size < minsize:
continue
ids.add(max_accn)
newids = remove_isoforms(ids)
logging.debug("Remove isoforms: before={0} after={1}".\
format(len(ids), len(newids)))
longestidsfile = pf + ".longest.ids"
fw = open(longestidsfile, "w")
print >> fw, "\n".join(newids)
fw.close()
logging.debug("A total of {0} records written to `{1}`.".\
format(len(newids), longestidsfile))
longestbedfile = pf + ".longest.bed"
some([
bedfile, longestidsfile, "--outfile={0}".format(longestbedfile),
"--no_strip_names"
])
def longest(args):
"""
%prog longest bedfile fastafile
Select longest feature within overlapping piles.
"""
from jcvi.formats.sizes import Sizes
p = OptionParser(longest.__doc__)
p.add_option("--maxsize", default=20000, type="int",
help="Limit max size")
p.add_option("--minsize", default=60, type="int",
help="Limit min size")
p.add_option("--precedence", default="Medtr",
help="Accessions with prefix take precedence")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
bedfile, fastafile = args
maxsize = opts.maxsize
minsize = opts.minsize
prec = opts.precedence
mergedbed = mergeBed(bedfile, nms=True)
sizes = Sizes(fastafile).mapping
bed = Bed(mergedbed)
pf = bedfile.rsplit(".", 1)[0]
ids = set()
for b in bed:
accns = b.accn.split(";")
prec_accns = [x for x in accns if x.startswith(prec)]
if prec_accns:
accns = prec_accns
accn_sizes = [(sizes.get(x, 0), x) for x in accns]
accn_sizes = [(size, x) for size, x in accn_sizes if size < maxsize]
if not accn_sizes:
continue
max_size, max_accn = max(accn_sizes)
if max_size < minsize:
continue
ids.add(max_accn)
newids = remove_isoforms(ids)
logging.debug("Remove isoforms: before={0} after={1}".\
format(len(ids), len(newids)))
longestidsfile = pf + ".longest.ids"
fw = open(longestidsfile, "w")
print >> fw, "\n".join(newids)
fw.close()
logging.debug("A total of {0} records written to `{1}`.".\
format(len(newids), longestidsfile))
longestbedfile = pf + ".longest.bed"
some([bedfile, longestidsfile, "--outfile={0}".format(longestbedfile),
"--no_strip_names"])
