def write_agp(self, filename):
sizes = self.sz
agp = []
for scaffold, lines in self.iter_scaffold():
for a, b in pairwise(lines):
cline = AGPLine.cline(scaffold, a.tig, sizes, a.oo)
gline = AGPLine.gline(scaffold, a.gaps)
agp.append(cline)
agp.append(gline)
a = lines[-1]
cline = AGPLine.cline(scaffold, a.tig, sizes, a.oo)
agp.append(cline)
fw = open(filename, "w")
for a in agp:
print >> fw, a
fw.close()
reindex([filename, "--inplace"])
return filename
def write_agp(self, filename):
sizes = self.sz
agp = []
for scaffold, lines in self.iter_scaffold():
for a, b in pairwise(lines):
cline = AGPLine.cline(scaffold, a.tig, sizes, a.oo)
gline = AGPLine.gline(scaffold, a.gaps)
agp.append(cline)
agp.append(gline)
a = lines[-1]
cline = AGPLine.cline(scaffold, a.tig, sizes, a.oo)
agp.append(cline)
fw = open(filename, "w")
for a in agp:
print >> fw, a
fw.close()
reindex([filename, "--inplace"])
return filename
def write_AGP(self, filename, orientationguide={}, reindex=True):
"""
For each component, we have two overlaps: North and South.
=======
|||| South
====(=================) Current BAC
North ||||
===============
For the case that says "Non-terminal", the overlap will not be
considered. North-South would suggest a '+' orientation, South-North
would suggest a '-' orientation. In most cases, unless the overlap
involves phase1 BAC, the selected range will be shown as the brackets
above - exclude North overlap, and include South overlap (aka the
"left-greedy" rule).
"""
fw = must_open(filename, "w")
for aid, bb in groupby(self.lines, key=lambda x: x.aid):
bb = list(bb)
north, south = bb
aid = north.aid
assert aid == south.aid
aphase = north.aphase
chr = north.chr
size = north.asize
ar = [chr, 0, 0, 0]
northline = southline = None
northrange = southrange = None
# Warn if adjacent components do not have valid # overlaps
if south.is_no_overlap:
print >> sys.stderr, south
# Most gaps, except telomeres occur twice, so only do the "North"
if north.is_gap:
bar = ar + self.get_agp_gap(north.bid)
northline = "\t".join(str(x) for x in bar)
else:
if north.isTerminal():
northrange = north.astart, north.astop
if south.is_gap:
if south.bid == "telomere":
bar = ar + self.get_agp_gap(south.bid)
southline = "\t".join(str(x) for x in bar)
else:
if south.isTerminal():
southrange = south.astart, south.astop
else:
bar = ar + self.get_agp_gap("fragment")
southline = "\t".join(str(x) for x in bar)
# Determine the orientation and clear range for the current BAC
clr = [1, size]
orientation = sorientation = None
if northrange:
start, stop = northrange
Lhang = start - 1
Rhang = size - stop
orientation = '+' if Lhang < Rhang else '-'
if north.bphase == 1 and north.bphase < aphase:
if Lhang < Rhang: # North overlap at 5`
clr[0] = start
else:
clr[1] = stop
# Override left-greedy (also see below)
else:
if Lhang < Rhang:
clr[0] = stop + 1
else:
clr[1] = start - 1
if southrange:
start, stop = southrange
Lhang = start - 1
Rhang = size - stop
sorientation = '+' if Lhang > Rhang else '-'
# Override left-greedy (also see above)
if aphase == 1 and aphase < south.bphase:
if Lhang < Rhang: # South overlap at 5`
clr[0] = stop + 1
else:
clr[1] = start - 1
else:
if Lhang < Rhang:
clr[0] = start
else:
clr[1] = stop
if orientation:
if sorientation:
try:
assert orientation == sorientation, \
"Orientation conflicts:\n{0}\n{1}".format(north, south)
except AssertionError as e:
logging.debug(e)
else:
if sorientation:
orientation = sorientation
else: # Both overlaps fail to define orientation
orientation = orientationguide.get(aid, "+")
component_type = "D" if aphase in (1, 2) else "F"
bar = ar + [component_type, aid, clr[0], clr[1], orientation]
cline = "\t".join(str(x) for x in bar)
if northline:
print >> fw, northline
print >> fw, cline
if southline:
print >> fw, southline
fw.close()
if reindex:
from jcvi.formats.agp import reindex
reindex([filename])
newagpfile = filename.replace(".agp", ".reindexed.agp")
shutil.move(newagpfile, filename)
def write_AGP(self, filename, orientationguide={}):
"""
For each component, we have two overlaps: North and South.
=======
|||| South
====(=================) Current BAC
North ||||
===============
For the case that says "Non-terminal", the overlap will not be
considered. North-South would suggest a '+' orientation, South-North
would suggest a '-' orientation. In most cases, unless the overlap
involves phase1 BAC, the selected range will be shown as the brackets
above - exclude North overlap, and include South overlap (aka the
"left-greedy" rule).
"""
fw = must_open(filename, "w")
for aid, bb in groupby(self.lines, key=lambda x: x.aid):
bb = list(bb)
north, south = bb
aid = north.aid
assert aid == south.aid
aphase = north.aphase
chr = north.chr
size = north.asize
ar = [chr, 0, 0, 0]
northline = southline = None
northrange = southrange = None
# Warn if adjacent components do not have valid overlaps
if south.is_no_overlap:
print >> sys.stderr, south
# Most gaps, except telomeres occur twice, so only do the "North"
if north.is_gap:
bar = ar + self.get_agp_gap(north.bid)
northline = "\t".join(str(x) for x in bar)
else:
if north.isTerminal:
northrange = north.astart, north.astop
if south.is_gap:
if south.bid == "telomere":
bar = ar + self.get_agp_gap(south.bid)
southline = "\t".join(str(x) for x in bar)
else:
if south.isTerminal:
southrange = south.astart, south.astop
else:
bar = ar + self.get_agp_gap("fragment")
southline = "\t".join(str(x) for x in bar)
# Determine the orientation and clear range for the current BAC
clr = [1, size]
orientation = sorientation = None
if northrange:
start, stop = northrange
Lhang = start - 1
Rhang = size - stop
orientation = '+' if Lhang < Rhang else '-'
if north.bphase == 1 and north.bphase < aphase:
if Lhang < Rhang: # North overlap at 5`
clr[0] = start
else:
clr[1] = stop
# Override left-greedy (also see below)
else:
if Lhang < Rhang:
clr[0] = stop + 1
else:
clr[1] = start - 1
if southrange:
start, stop = southrange
Lhang = start - 1
Rhang = size - stop
sorientation = '+' if Lhang > Rhang else '-'
# Override left-greedy (also see above)
if aphase == 1 and aphase < south.bphase:
if Lhang < Rhang: # South overlap at 5`
clr[0] = stop + 1
else:
clr[1] = start - 1
else:
if Lhang < Rhang:
clr[0] = start
else:
clr[1] = stop
if orientation:
if sorientation:
try:
assert orientation == sorientation, \
"Orientation conflicts:\n{0}\n{1}".format(north, south)
except AssertionError as e:
logging.debug(e)
else:
if sorientation:
orientation = sorientation
else: # Both overlaps fail to define orientation
orientation = orientationguide.get(aid, "+")
component_type = "D" if aphase in (1, 2) else "F"
bar = ar + [component_type, aid, clr[0], clr[1], orientation]
cline = "\t".join(str(x) for x in bar)
if northline:
print >> fw, northline
print >> fw, cline
if southline:
print >> fw, southline
fw.close()
reindex([filename, "--inplace"])
def estimategaps(args):
"""
%prog estimategaps input.bed
Estimate sizes of inter-scaffold gaps. The AGP file generated by path()
command has unknown gap sizes with a generic number of Ns (often 100 Ns).
The AGP file `input.chr.agp` will be modified in-place.
"""
p = OptionParser(estimategaps.__doc__)
p.add_option("--minsize", default=100, type="int", help="Minimum gap size")
p.add_option("--maxsize",
default=500000,
type="int",
help="Maximum gap size")
p.add_option("--links",
default=10,
type="int",
help="Only use linkage grounds with matchings more than")
p.set_verbose(help="Print details for each gap calculation")
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
inputbed, = args
pf = inputbed.rsplit(".", 1)[0]
agpfile = pf + ".chr.agp"
bedfile = pf + ".lifted.bed"
cc = Map(bedfile, scaffold_info=True)
agp = AGP(agpfile)
minsize, maxsize = opts.minsize, opts.maxsize
links = opts.links
verbose = opts.verbose
outagpfile = pf + ".estimategaps.agp"
fw = must_open(outagpfile, "w")
for ob, components in agp.iter_object():
components = list(components)
s = Scaffold(ob, cc)
mlg_counts = s.mlg_counts
gaps = [x for x in components if x.is_gap]
gapsizes = [None] * len(gaps) # master
for mlg, count in mlg_counts.items():
if count < links:
continue
g = GapEstimator(cc, agp, ob, mlg)
g.compute_all_gaps(minsize=minsize, maxsize=maxsize, \
verbose=verbose)
# Merge evidence from this mlg into master
assert len(g.gapsizes) == len(gaps)
for i, gs in enumerate(gapsizes):
gg = g.gapsizes[i]
if gs is None:
gapsizes[i] = gg
elif gg:
gapsizes[i] = min(gs, gg)
print gapsizes
# Modify AGP
i = 0
for x in components:
if x.is_gap:
x.gap_length = gapsizes[i] or minsize
x.component_type = 'U' if x.gap_length == 100 else 'N'
i += 1
print >> fw, x
fw.close()
reindex([outagpfile, "--inplace"])
def estimategaps(args):
"""
%prog estimategaps input.bed
Estimate sizes of inter-scaffold gaps. The AGP file generated by path()
command has unknown gap sizes with a generic number of Ns (often 100 Ns).
The AGP file `input.chr.agp` will be modified in-place.
"""
p = OptionParser(estimategaps.__doc__)
p.add_option("--minsize", default=100, type="int",
help="Minimum gap size")
p.add_option("--maxsize", default=500000, type="int",
help="Maximum gap size")
p.add_option("--links", default=10, type="int",
help="Only use linkage grounds with matchings more than")
p.set_verbose(help="Print details for each gap calculation")
opts, args = p.parse_args(args)
if len(args) != 1:
sys.exit(not p.print_help())
inputbed, = args
pf = inputbed.rsplit(".", 1)[0]
agpfile = pf + ".chr.agp"
bedfile = pf + ".lifted.bed"
cc = Map(bedfile, scaffold_info=True)
agp = AGP(agpfile)
minsize, maxsize = opts.minsize, opts.maxsize
links = opts.links
verbose = opts.verbose
outagpfile = pf + ".estimategaps.agp"
fw = must_open(outagpfile, "w")
for ob, components in agp.iter_object():
components = list(components)
s = Scaffold(ob, cc)
mlg_counts = s.mlg_counts
gaps = [x for x in components if x.is_gap]
gapsizes = [None] * len(gaps) # master
for mlg, count in mlg_counts.items():
if count < links:
continue
g = GapEstimator(cc, agp, ob, mlg)
g.compute_all_gaps(minsize=minsize, maxsize=maxsize, \
verbose=verbose)
# Merge evidence from this mlg into master
assert len(g.gapsizes) == len(gaps)
for i, gs in enumerate(gapsizes):
gg = g.gapsizes[i]
if gs is None:
gapsizes[i] = gg
elif gg:
gapsizes[i] = min(gs, gg)
print gapsizes
# Modify AGP
i = 0
for x in components:
if x.is_gap:
x.gap_length = gapsizes[i] or minsize
x.component_type = 'U' if x.gap_length == 100 else 'N'
i += 1
print >> fw, x
fw.close()
reindex([outagpfile, "--inplace"])