class OVL (LineFile):
def __init__(self, filename):
super(OVL, self).__init__(filename)
fp = must_open(filename)
contained = set()
for row in fp:
o = OVLLine(row)
self.append(o)
if o.tag == "a in b":
contained.add(o.a)
elif o.tag == "b in a":
contained.add(o.b)
logging.debug("Imported {} links. Contained tigs: {}".\
format(len(self), len(contained)))
self.contained = contained
self.graph = BiGraph()
for o in self:
if o.tag == "a->b":
a, b = o.a, o.b
elif o.tag == "b->a":
a, b = o.b, o.a
if a in contained or b in contained:
continue
bstrand = '<' if o.bstrand == '-' else '>'
self.graph.add_edge(a, b, '>', bstrand, length=o.score)
def partition(args):
"""
%prog partition happy.txt synteny.graph
Select edges from another graph and merge it with the certain edges built
from the HAPPY mapping data.
"""
allowed_format = ("png", "ps")
p = OptionParser(partition.__doc__)
p.add_option("--prefix", help="Add prefix to the name [default: %default]")
p.add_option("--namestart", default=0, type="int",
help="Use a shorter name, starting index [default: %default]")
p.add_option("--format", default="png", choices=allowed_format,
help="Generate image of format [default: %default]")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
happyfile, graphfile = args
bg = BiGraph()
bg.read(graphfile, color="red")
prefix = opts.prefix
fp = open(happyfile)
for i, row in enumerate(fp):
nns = happy_nodes(row, prefix=prefix)
nodes = set(nns)
edges = happy_edges(row, prefix=prefix)
small_graph = BiGraph()
for e, is_uncertain in edges:
if is_uncertain:
e.color = "gray"
small_graph.add_edge(e)
for (u, v), e in bg.edges.items():
# Grab edge if both vertices are on the same line
if u in nodes and v in nodes:
uv = (str(u), str(v))
if uv in small_graph.edges:
e = small_graph.edges[uv]
e.color = "blue" # supported by both evidences
else:
small_graph.add_edge(e)
print >> sys.stderr, small_graph
pngfile = "A{0:02d}.{1}".format(i + 1, opts.format)
telomeres = (nns[0], nns[-1])
small_graph.draw(pngfile, namestart=opts.namestart,
nodehighlight=telomeres, dpi=72)
legend = ["Edge colors:"]
legend.append("[BLUE] Experimental + Synteny")
legend.append("[BLACK] Experimental certain")
legend.append("[GRAY] Experimental uncertain")
legend.append("[RED] Synteny only")
legend.append("Rectangle nodes are telomeres.")
print >> sys.stderr, "\n".join(legend)
def fromblast(args):
"""
%prog fromblast blastfile subject.fasta
Generate path from BLAST file. If multiple subjects map to the same query,
an edge is constructed between them (with the link provided by the query).
The BLAST file MUST be filtered, chained, supermapped.
"""
from jcvi.formats.blast import sort
from jcvi.utils.range import range_distance
p = OptionParser(fromblast.__doc__)
p.add_option(
"--clique",
default=False,
action="store_true",
help="Populate clique instead of linear path",
)
p.add_option(
"--maxdist",
default=100000,
type="int",
help="Create edge within certain distance",
)
p.set_verbose(help="Print verbose reports to stdout")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
blastfile, subjectfasta = args
clique = opts.clique
maxdist = opts.maxdist
sort([blastfile, "--query"])
blast = BlastSlow(blastfile, sorted=True)
g = BiGraph()
for query, blines in groupby(blast, key=lambda x: x.query):
blines = list(blines)
iterator = combinations(blines, 2) if clique else pairwise(blines)
for a, b in iterator:
asub, bsub = a.subject, b.subject
if asub == bsub:
continue
arange = (a.query, a.qstart, a.qstop, "+")
brange = (b.query, b.qstart, b.qstop, "+")
dist, oo = range_distance(arange, brange, distmode="ee")
if dist > maxdist:
continue
atag = ">" if a.orientation == "+" else "<"
btag = ">" if b.orientation == "+" else "<"
g.add_edge(asub, bsub, atag, btag)
graph_to_agp(g, blastfile, subjectfasta, verbose=opts.verbose)
def graph(self):
g = BiGraph()
for scaffold, lines in self.iter_scaffold():
self.scf[scaffold] = [x.tig for x in lines]
for a, b in pairwise(lines):
g.add_edge(a.tig, b.tig, a.o, b.o, length=a.gaps)
if len(lines) == 1: # Singleton scaffold
a = lines[0]
g.add_node(a.tig)
return g
def graph(self):
g = BiGraph()
for scaffold, lines in self.iter_scaffold():
self.scf[scaffold] = [x.tig for x in lines]
for a, b in pairwise(lines):
g.add_edge(a.tig, b.tig, a.o, b.o, length=a.gaps)
if len(lines) == 1: # Singleton scaffold
a = lines[0]
g.add_node(a.tig)
return g
def fromblast(args):
"""
%prog fromblast blastfile subject.fasta
Generate path from BLAST file. If multiple subjects map to the same query,
an edge is constructed between them (with the link provided by the query).
The BLAST file MUST be filtered, chained, supermapped.
"""
from jcvi.formats.blast import sort
from jcvi.utils.range import range_distance
p = OptionParser(fromblast.__doc__)
p.add_option("--clique", default=False, action="store_true",
help="Populate clique instead of linear path [default: %default]")
p.add_option("--maxdist", default=100000, type="int",
help="Create edge within certain distance [default: %default]")
p.add_option("--verbose", default=False, action="store_true",
help="Print verbose reports to stdout [default: %default]")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
blastfile, subjectfasta = args
clique = opts.clique
maxdist = opts.maxdist
sort([blastfile, "--query"])
blast = BlastSlow(blastfile, sorted=True)
g = BiGraph()
for query, blines in groupby(blast, key=lambda x: x.query):
blines = list(blines)
iterator = combinations(blines, 2) if clique else pairwise(blines)
for a, b in iterator:
asub, bsub = a.subject, b.subject
if asub == bsub:
continue
arange = (a.query, a.qstart, a.qstop, "+")
brange = (b.query, b.qstart, b.qstop, "+")
dist, oo = range_distance(arange, brange, distmode="ee")
if dist > maxdist:
continue
atag = ">" if a.orientation == "+" else "<"
btag = ">" if b.orientation == "+" else "<"
g.add_edge(BiEdge(asub, bsub, atag, btag))
graph_to_agp(g, blastfile, subjectfasta, verbose=opts.verbose)
def graph(args):
"""
%prog graph all.ov contigs.fasta
Load overlaps to create bidirectional graph.
"""
from jcvi.algorithms.graph import BiGraph, BiEdge
from jcvi.assembly.syntenypath import graph_to_agp
p = OptionParser(graph.__doc__)
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
allov, contigsfasta = args
g = BiGraph()
fp = open(allov)
# ctg7180000045865 - ctg7180000086408: a ~ b Overlap: 5108 Identity: 99.02%
# Orientation: +
contained = set()
for row in fp:
atoms = row.strip().split(":")
pair = atoms[0]
a, b = pair.split(" - ")
orientation = atoms[-1].strip()
tag = atoms[1].replace("Overlap", "").strip()
oa = ">"
ob = "<" if orientation == '-' else ">"
if tag == "a ~ b":
g.add_edge(BiEdge(a, b, oa, ob))
elif tag == "b ~ a":
g.add_edge(BiEdge(b, a, ob, oa))
elif tag == "a in b":
contained.add(a)
elif tag == "b in a":
contained.add(b)
graph_to_agp(g, allov, contigsfasta)
containedfile = "contained.ids"
fw = open(containedfile, "w")
print >> fw, "\n".join(contained)
fw.close()
class OVL(LineFile):
def __init__(self, filename):
super(OVL, self).__init__(filename)
fp = must_open(filename)
contained = set()
alledges = defaultdict(list)
for row in fp:
o = OVLLine(row)
self.append(o)
if o.tag == "a in b":
contained.add(o.a)
elif o.tag == "b in a":
contained.add(o.b)
if o.tag == "a->b":
alledges[o.a + "-3`"].append(o)
elif o.tag == "b->a":
alledges[o.a + "-5`"].append(o)
logging.debug(
"Imported {} links. Contained tigs: {}".format(len(self), len(contained))
)
self.contained = contained
logging.debug("Pruning edges to keep the mutual best")
for k, v in alledges.items():
bo = max(v, key=lambda x: x.score)
bo.best = True
self.graph = BiGraph()
for o in self:
if not o.best:
continue
if o.tag == "a->b":
a, b = o.a, o.b
elif o.tag == "b->a":
a, b = o.b, o.a
if a in contained or b in contained:
continue
bstrand = "<" if o.bstrand == "-" else ">"
self.graph.add_edge(a, b, ">", bstrand, length=o.score)
class OVL (LineFile):
def __init__(self, filename):
super(OVL, self).__init__(filename)
fp = must_open(filename)
contained = set()
alledges = defaultdict(list)
for row in fp:
o = OVLLine(row)
self.append(o)
if o.tag == "a in b":
contained.add(o.a)
elif o.tag == "b in a":
contained.add(o.b)
if o.tag == "a->b":
alledges[o.a + "-3`"].append(o)
elif o.tag == "b->a":
alledges[o.a + "-5`"].append(o)
logging.debug("Imported {} links. Contained tigs: {}".\
format(len(self), len(contained)))
self.contained = contained
logging.debug("Pruning edges to keep the mutual best")
for k, v in alledges.items():
bo = max(v, key=lambda x: x.score)
bo.best = True
self.graph = BiGraph()
for o in self:
if not o.best:
continue
if o.tag == "a->b":
a, b = o.a, o.b
elif o.tag == "b->a":
a, b = o.b, o.a
if a in contained or b in contained:
continue
bstrand = '<' if o.bstrand == '-' else '>'
self.graph.add_edge(a, b, '>', bstrand, length=o.score)
def connect(args):
"""
%prog connect assembly.fasta read_mapping.blast
Connect contigs using long reads.
"""
p = OptionParser(connect.__doc__)
p.add_option(
"--clip",
default=2000,
type="int",
help="Only consider end of contigs",
)
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
fastafile, blastfile = args
clip = opts.clip
sizes = Sizes(fastafile).mapping
blast = Blast(blastfile)
blasts = []
for b in blast:
seqid = b.subject
size = sizes[seqid]
start, end = b.sstart, b.sstop
cstart, cend = min(size, clip), max(0, size - clip)
if start > cstart and end < cend:
continue
blasts.append(b)
key = lambda x: x.query
blasts.sort(key=key)
g = BiGraph()
for query, bb in groupby(blasts, key=key):
bb = sorted(bb, key=lambda x: x.qstart)
nsubjects = len(set(x.subject for x in bb))
if nsubjects == 1:
continue
print("\n".join(str(x) for x in bb))
for a, b in pairwise(bb):
astart, astop = a.qstart, a.qstop
bstart, bstop = b.qstart, b.qstop
if a.subject == b.subject:
continue
arange = astart, astop
brange = bstart, bstop
ov = range_intersect(arange, brange)
alen = astop - astart + 1
blen = bstop - bstart + 1
if ov:
ostart, ostop = ov
ov = ostop - ostart + 1
print(ov, alen, blen)
if ov and (ov > alen / 2 or ov > blen / 2):
print("Too much overlap ({0})".format(ov))
continue
asub = a.subject
bsub = b.subject
atag = ">" if a.orientation == "+" else "<"
btag = ">" if b.orientation == "+" else "<"
g.add_edge(asub, bsub, atag, btag)
graph_to_agp(g, blastfile, fastafile, verbose=False)
def partition(args):
"""
%prog partition happy.txt synteny.graph
Select edges from another graph and merge it with the certain edges built
from the HAPPY mapping data.
"""
allowed_format = ("png", "ps")
p = OptionParser(partition.__doc__)
p.add_option("--prefix", help="Add prefix to the name")
p.add_option(
"--namestart",
default=0,
type="int",
help="Use a shorter name, starting index",
)
p.add_option(
"--format",
default="png",
choices=allowed_format,
help="Generate image of format",
)
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
happyfile, graphfile = args
bg = BiGraph()
bg.read(graphfile, color="red")
prefix = opts.prefix
fp = open(happyfile)
for i, row in enumerate(fp):
nns = happy_nodes(row, prefix=prefix)
nodes = set(nns)
edges = happy_edges(row, prefix=prefix)
small_graph = BiGraph()
for (a, b, oa, ob), is_uncertain in edges:
color = "gray" if is_uncertain else "black"
small_graph.add_edge(a, b, oa, ob, color=color)
for (u, v), e in bg.edges.items():
# Grab edge if both vertices are on the same line
if u in nodes and v in nodes:
uv = (str(u), str(v))
if uv in small_graph.edges:
e = small_graph.edges[uv]
e.color = "blue" # supported by both evidences
else:
small_graph.add_edge(e)
print(small_graph, file=sys.stderr)
pngfile = "A{0:02d}.{1}".format(i + 1, opts.format)
telomeres = (nns[0], nns[-1])
small_graph.draw(
pngfile, namestart=opts.namestart, nodehighlight=telomeres, dpi=72
)
legend = [
"Edge colors:",
"[BLUE] Experimental + Synteny",
"[BLACK] Experimental certain",
"[GRAY] Experimental uncertain",
"[RED] Synteny only",
"Rectangle nodes are telomeres.",
]
print("\n".join(legend), file=sys.stderr)
def connect(args):
"""
%prog connect assembly.fasta read_mapping.blast
Connect contigs using long reads.
"""
from jcvi.formats.sizes import Sizes
from jcvi.formats.blast import Blast
from jcvi.utils.iter import pairwise
from jcvi.utils.range import range_intersect
from jcvi.algorithms.graph import BiGraph, BiEdge
from jcvi.assembly.syntenypath import graph_to_agp
p = OptionParser(connect.__doc__)
p.add_option("--clip", default=2000, type="int",
help="Only consider end of contigs [default: %default]")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
fastafile, blastfile = args
clip = opts.clip
sizes = Sizes(fastafile).mapping
blast = Blast(blastfile)
blasts = []
for b in blast:
seqid = b.subject
size = sizes[seqid]
start, end = b.sstart, b.sstop
cstart, cend = min(size, clip), max(0, size - clip)
if start > cstart and end < cend:
continue
blasts.append(b)
key = lambda x: x.query
blasts.sort(key=key)
g = BiGraph()
for query, bb in groupby(blasts, key=key):
bb = sorted(bb, key=lambda x: x.qstart)
nsubjects = len(set(x.subject for x in bb))
if nsubjects == 1:
continue
print "\n".join(str(x) for x in bb)
for a, b in pairwise(bb):
astart, astop = a.qstart, a.qstop
bstart, bstop = b.qstart, b.qstop
if a.subject == b.subject:
continue
arange = astart, astop
brange = bstart, bstop
ov = range_intersect(arange, brange)
alen = astop - astart + 1
blen = bstop - bstart + 1
if ov:
ostart, ostop = ov
ov = ostop - ostart + 1
print ov, alen, blen
if ov and (ov > alen / 2 or ov > blen / 2):
print "Too much overlap ({0})".format(ov)
continue
asub = a.subject
bsub = b.subject
atag = ">" if a.orientation == "+" else "<"
btag = ">" if b.orientation == "+" else "<"
e = BiEdge(asub, bsub, atag, btag)
g.add_edge(e)
print "=" * 5, e
graph_to_agp(g, blastfile, fastafile, verbose=False)
def fromblast(args):
"""
%prog fromblast blastfile subject.fasta
Generate path from BLAST file. If multiple subjects map to the same query,
an edge is constructed between them (with the link provided by the query).
The BLAST file MUST be filtered, chained, supermapped.
"""
from jcvi.formats.blast import sort
from jcvi.utils.range import range_distance
p = OptionParser(fromblast.__doc__)
p.add_option(
"--clique",
default=False,
action="store_true",
help="Populate clique instead of linear path [default: %default]")
p.add_option(
"--maxdist",
default=100000,
type="int",
help="Create edge within certain distance [default: %default]")
p.add_option("--verbose",
default=False,
action="store_true",
help="Print verbose reports to stdout [default: %default]")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
blastfile, subjectfasta = args
clique = opts.clique
maxdist = opts.maxdist
sort([blastfile, "--query"])
blast = BlastSlow(blastfile, sorted=True)
g = BiGraph()
for query, blines in groupby(blast, key=lambda x: x.query):
blines = list(blines)
iterator = combinations(blines, 2) if clique else pairwise(blines)
for a, b in iterator:
asub, bsub = a.subject, b.subject
if asub == bsub:
continue
arange = (a.query, a.qstart, a.qstop, "+")
brange = (b.query, b.qstart, b.qstop, "+")
dist, oo = range_distance(arange, brange, distmode="ee")
if dist > maxdist:
continue
atag = ">" if a.orientation == "+" else "<"
btag = ">" if b.orientation == "+" else "<"
g.add_edge(BiEdge(asub, bsub, atag, btag))
g.write("graph.txt")
#g.draw("graph.pdf")
logging.debug(str(g))
paths = []
for path in g.iter_paths():
m, oo = g.path(path)
if len(oo) == 1: # Singleton path
continue
paths.append(oo)
if opts.verbose:
print m
print oo
npaths = len(paths)
ntigs = sum(len(x) for x in paths)
logging.debug("Graph decomposed to {0} paths with {1} components.".\
format(npaths, ntigs))
agpfile = blastfile + ".agp"
sizes = Sizes(subjectfasta)
fwagp = open(agpfile, "w")
scaffolded = set()
for i, oo in enumerate(paths):
ctgorder = [(str(ctg), ("+" if strand else "-")) \
for ctg, strand in oo]
scaffolded |= set(ctg for ctg, strand in ctgorder)
object = "pmol_{0:04d}".format(i)
order_to_agp(object, ctgorder, sizes.mapping, fwagp)
# Get the singletons as well
nsingletons = 0
for ctg, size in sizes.iter_sizes():
if ctg in scaffolded:
continue
ctgorder = [(ctg, "+")]
object = ctg
order_to_agp(object, ctgorder, sizes.mapping, fwagp)
nsingletons += 1
logging.debug("Written {0} unscaffolded singletons.".format(nsingletons))
fwagp.close()
logging.debug("AGP file written to `{0}`.".format(agpfile))
def fromblast(args):
"""
%prog fromblast blastfile subject.fasta
Generate path from BLAST file. If multiple subjects map to the same query,
an edge is constructed between them (with the link provided by the query).
The BLAST file MUST be filtered, chained, supermapped.
"""
from jcvi.formats.blast import sort
from jcvi.utils.range import range_distance
p = OptionParser(fromblast.__doc__)
p.add_option("--clique", default=False, action="store_true",
help="Populate clique instead of linear path [default: %default]")
p.add_option("--maxdist", default=100000, type="int",
help="Create edge within certain distance [default: %default]")
p.add_option("--verbose", default=False, action="store_true",
help="Print verbose reports to stdout [default: %default]")
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(not p.print_help())
blastfile, subjectfasta = args
clique = opts.clique
maxdist = opts.maxdist
sort([blastfile, "--query"])
blast = BlastSlow(blastfile, sorted=True)
g = BiGraph()
for query, blines in groupby(blast, key=lambda x: x.query):
blines = list(blines)
iterator = combinations(blines, 2) if clique else pairwise(blines)
for a, b in iterator:
asub, bsub = a.subject, b.subject
if asub == bsub:
continue
arange = (a.query, a.qstart, a.qstop, "+")
brange = (b.query, b.qstart, b.qstop, "+")
dist, oo = range_distance(arange, brange, distmode="ee")
if dist > maxdist:
continue
atag = ">" if a.orientation == "+" else "<"
btag = ">" if b.orientation == "+" else "<"
g.add_edge(BiEdge(asub, bsub, atag, btag))
g.write("graph.txt")
#g.draw("graph.pdf")
logging.debug(str(g))
paths = []
for path in g.iter_paths():
m, oo = g.path(path)
if len(oo) == 1: # Singleton path
continue
paths.append(oo)
if opts.verbose:
print m
print oo
npaths = len(paths)
ntigs = sum(len(x) for x in paths)
logging.debug("Graph decomposed to {0} paths with {1} components.".\
format(npaths, ntigs))
agpfile = blastfile + ".agp"
sizes = Sizes(subjectfasta)
fwagp = open(agpfile, "w")
scaffolded = set()
for i, oo in enumerate(paths):
ctgorder = [(str(ctg), ("+" if strand else "-")) \
for ctg, strand in oo]
scaffolded |= set(ctg for ctg, strand in ctgorder)
object = "pmol_{0:04d}".format(i)
order_to_agp(object, ctgorder, sizes.mapping, fwagp)
# Get the singletons as well
nsingletons = 0
for ctg, size in sizes.iter_sizes():
if ctg in scaffolded:
continue
ctgorder = [(ctg, "+")]
object = ctg
order_to_agp(object, ctgorder, sizes.mapping, fwagp)
nsingletons += 1
logging.debug("Written {0} unscaffolded singletons.".format(nsingletons))
fwagp.close()
logging.debug("AGP file written to `{0}`.".format(agpfile))