def filter(args):
"""
%prog filter <deltafile|coordsfile>
Produce a new delta/coords file and filter based on id% or cov%.
Use `delta-filter` for .delta file.
"""
p = OptionParser(filter.__doc__)
p.set_align(pctid=0, hitlen=0)
p.add_option("--overlap", default=False, action="store_true",
help="Print overlap status (e.g. terminal, contained)")
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
pctid = opts.pctid
hitlen = opts.hitlen
filename, = args
if pctid == 0 and hitlen == 0:
return filename
pf, suffix = filename.rsplit(".", 1)
outfile = "".join((pf, ".P{0}L{1}.".format(pctid, hitlen), suffix))
if not need_update(filename, outfile):
return outfile
if suffix == "delta":
cmd = "delta-filter -i {0} -l {1} {2}".format(pctid, hitlen, filename)
sh(cmd, outfile=outfile)
return outfile
fp = open(filename)
fw = must_open(outfile, "w")
for row in fp:
try:
c = CoordsLine(row)
except AssertionError:
continue
if c.identity < pctid:
continue
if c.len2 < hitlen:
continue
if opts.overlap and not c.overlap:
continue
outrow = row.rstrip()
if opts.overlap:
ov = Overlap_types[c.overlap]
outrow += "\t" + ov
print >> fw, outrow
return outfile
def __init__(self, filename, sorted=False, header=False):
if filename.endswith(".delta"):
coordsfile = filename.rsplit(".", 1)[0] + ".coords"
if need_update(filename, coordsfile):
fromdelta([filename])
filename = coordsfile
super(Coords, self).__init__(filename)
fp = open(filename)
if header:
self.cmd = fp.next()
for row in fp:
try:
self.append(CoordsLine(row))
except AssertionError, e:
pass
def blast(args):
"""
%prog blast <deltafile|coordsfile>
Covert delta or coordsfile to BLAST tabular output.
"""
p = OptionParser(blast.__doc__)
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
deltafile, = args
blastfile = deltafile.rsplit(".", 1)[0] + ".blast"
if need_update(deltafile, blastfile):
coords = Coords(deltafile)
fw = open(blastfile, "w")
for c in coords:
print >> fw, c.blastline
