def clr(args):
"""
%prog blastfile fastafiles
Calculate the vector clear range file based BLAST to the vectors.
"""
p = OptionParser(clr.__doc__)
opts, args = p.parse_args(args)
if len(args) < 2:
sys.exit(not p.print_help())
blastfile = args[0]
fastafiles = args[1:]
sizes = {}
for fa in fastafiles:
f = Fasta(fa)
sizes.update(f.itersizes())
b = Blast(blastfile)
seen = set()
for query, hits in b.iter_hits():
qsize = sizes[query]
vectors = list((x.qstart, x.qstop) for x in hits)
vmin, vmax = range_minmax(vectors)
left_size = vmin - 1
right_size = qsize - vmax
if left_size > right_size:
clr_start, clr_end = 0, vmin
else:
clr_start, clr_end = vmax, qsize
print "\t".join(str(x) for x in (query, clr_start, clr_end))
del sizes[query]
for q, size in sorted(sizes.items()):
print "\t".join(str(x) for x in (q, 0, size))
def clr(args):
"""
%prog blastfile fastafiles
Calculate the vector clear range file based BLAST to the vectors.
"""
p = OptionParser(clr.__doc__)
opts, args = p.parse_args(args)
if len(args) < 2:
sys.exit(not p.print_help())
blastfile = args[0]
fastafiles = args[1:]
sizes = {}
for fa in fastafiles:
f = Fasta(fa)
sizes.update(f.itersizes())
b = Blast(blastfile)
seen = set()
for query, hits in b.iter_hits():
qsize = sizes[query]
vectors = list((x.qstart, x.qstop) for x in hits)
vmin, vmax = range_minmax(vectors)
left_size = vmin - 1
right_size = qsize - vmax
if left_size > right_size:
clr_start, clr_end = 0, vmin
else:
clr_start, clr_end = vmax, qsize
print "\t".join(str(x) for x in (query, clr_start, clr_end))
del sizes[query]
for q, size in sorted(sizes.items()):
print "\t".join(str(x) for x in (q, 0, size))
def loss(args):
"""
%prog loss a.b.i1.blocks [a.b-genomic.blast]
Extract likely gene loss candidates between genome a and b.
"""
p = OptionParser(loss.__doc__)
p.add_option("--bed", default=False, action="store_true",
help="Genomic BLAST is in bed format [default: %default]")
p.add_option("--gdist", default=20, type="int",
help="Gene distance [default: %default]")
p.add_option("--bdist", default=20000, type="int",
help="Base pair distance [default: %default]")
p.set_beds()
opts, args = p.parse_args(args)
if len(args) not in (1, 2):
sys.exit(not p.print_help())
blocksfile = args[0]
emptyblast = (len(args) == 1)
if emptyblast:
genomicblast = "empty.blast"
sh("touch {0}".format(genomicblast))
else:
genomicblast = args[1]
gdist, bdist = opts.gdist, opts.bdist
qbed, sbed, qorder, sorder, is_self = check_beds(blocksfile, p, opts)
blocks = []
fp = open(blocksfile)
genetrack = {}
proxytrack = {}
for row in fp:
a, b = row.split()
genetrack[a] = b
blocks.append((a, b))
data = []
for key, rows in groupby(blocks, key=lambda x: x[-1]):
rows = list(rows)
data.append((key, rows))
imax = len(data) - 1
for i, (key, rows) in enumerate(data):
if i == 0 or i == imax:
continue
if key != '.':
continue
before, br = data[i - 1]
after, ar = data[i + 1]
bi, bx = sorder[before]
ai, ax = sorder[after]
dist = abs(bi - ai)
if bx.seqid != ax.seqid or dist > gdist:
continue
start, end = range_minmax(((bx.start, bx.end), (ax.start, ax.end)))
start, end = max(start - bdist, 1), end + bdist
proxy = (bx.seqid, start, end)
for a, b in rows:
proxytrack[a] = proxy
tags = {}
if opts.bed:
bed = Bed(genomicblast, sorted=False)
key = lambda x: gene_name(x.accn.rsplit(".", 1)[0])
for query, bb in groupby(bed, key=key):
bb = list(bb)
if query not in proxytrack:
continue
proxy = proxytrack[query]
tag = "NS"
best_b = bb[0]
for b in bb:
hsp = (b.seqid, b.start, b.end)
if range_overlap(proxy, hsp):
tag = "S"
best_b = b
break
hsp = (best_b.seqid, best_b.start, best_b.end)
proxytrack[query] = hsp
tags[query] = tag
else:
blast = Blast(genomicblast)
for query, bb in blast.iter_hits():
bb = list(bb)
query = gene_name(query)
if query not in proxytrack:
continue
proxy = proxytrack[query]
tag = "NS"
best_b = bb[0]
for b in bb:
hsp = (b.subject, b.sstart, b.sstop)
if range_overlap(proxy, hsp):
tag = "S"
best_b = b
break
hsp = (best_b.subject, best_b.sstart, best_b.sstop)
proxytrack[query] = hsp
tags[query] = tag
for b in qbed:
accn = b.accn
target_region = genetrack[accn]
if accn in proxytrack:
target_region = region_str(proxytrack[accn])
if accn in tags:
ptag = "[{0}]".format(tags[accn])
else:
ptag = "[NF]"
target_region = ptag + target_region
print "\t".join((b.seqid, accn, target_region))
if emptyblast:
sh("rm -f {0}".format(genomicblast))
def rnaseq(args):
"""
%prog rnaseq blastfile ref.fasta
Evaluate de-novo RNA-seq assembly against a reference gene set (same or
closely related organism). Ideally blatfile needs to be supermap'd.
Following metric is used (Martin et al. 2010, Rnnotator paper):
Accuracy: % of contigs share >=95% identity with ref genome (TODO)
Completeness: % of ref genes covered by contigs to >=80% of their lengths
Contiguity: % of ref genes covered by a *single* contig >=80% of lengths
Chimer: % of contigs that contain two or more annotated genes >= 50bp
"""
from jcvi.algorithms.supermap import supermap
p = OptionParser(rnaseq.__doc__)
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(p.print_help())
blastfile, reffasta = args
sizes = Sizes(reffasta).mapping
known_genes = len(sizes)
querysupermap = blastfile + ".query.supermap"
refsupermap = blastfile + ".ref.supermap"
if not op.exists(querysupermap):
supermap(blastfile, filter="query")
if not op.exists(refsupermap):
supermap(blastfile, filter="ref")
blast = Blast(querysupermap)
chimers = 0
goodctg80 = set()
goodctg50 = set()
for ctg, hits in blast.iter_hits():
bps = defaultdict(int)
for x in hits:
bps[x.subject] += abs(x.sstop - x.sstart) + 1
valid_hits = bps.items()
for vh, length in valid_hits:
rsize = sizes[vh]
ratio = length * 100. / rsize
if ratio >= 80:
goodctg80.add(ctg)
if ratio >= 50:
goodctg50.add(ctg)
# Chimer
if len(valid_hits) > 1:
chimers += 1
blast = Blast(refsupermap)
goodref80 = set()
goodref50 = set()
bps = defaultdict(int)
for x in blast.iter_line():
bps[x.subject] += abs(x.sstop - x.sstart) + 1
for vh, length in bps.items():
rsize = sizes[vh]
ratio = length * 100. / rsize
if ratio >= 80:
goodref80.add(vh)
if ratio >= 50:
goodref50.add(vh)
print >> sys.stderr, "Reference set: `{0}`, # of transcripts {1}".\
format(reffasta, known_genes)
print >> sys.stderr, "A total of {0} contigs map to 80% of a reference"\
" transcript".format(len(goodctg80))
print >> sys.stderr, "A total of {0} contigs map to 50% of a reference"\
" transcript".format(len(goodctg50))
print >> sys.stderr, "A total of {0} reference transcripts ({1:.1f}%) have 80% covered" \
.format(len(goodref80), len(goodref80) * 100. / known_genes)
def loss(args):
"""
%prog loss a.b.i1.blocks [a.b-genomic.blast]
Extract likely gene loss candidates between genome a and b.
"""
p = OptionParser(loss.__doc__)
p.add_option("--bed",
default=False,
action="store_true",
help="Genomic BLAST is in bed format [default: %default]")
p.add_option("--gdist",
default=20,
type="int",
help="Gene distance [default: %default]")
p.add_option("--bdist",
default=20000,
type="int",
help="Base pair distance [default: %default]")
p.set_beds()
opts, args = p.parse_args(args)
if len(args) not in (1, 2):
sys.exit(not p.print_help())
blocksfile = args[0]
emptyblast = (len(args) == 1)
if emptyblast:
genomicblast = "empty.blast"
sh("touch {0}".format(genomicblast))
else:
genomicblast = args[1]
gdist, bdist = opts.gdist, opts.bdist
qbed, sbed, qorder, sorder, is_self = check_beds(blocksfile, p, opts)
blocks = []
fp = open(blocksfile)
genetrack = {}
proxytrack = {}
for row in fp:
a, b = row.split()
genetrack[a] = b
blocks.append((a, b))
data = []
for key, rows in groupby(blocks, key=lambda x: x[-1]):
rows = list(rows)
data.append((key, rows))
imax = len(data) - 1
for i, (key, rows) in enumerate(data):
if i == 0 or i == imax:
continue
if key != '.':
continue
before, br = data[i - 1]
after, ar = data[i + 1]
bi, bx = sorder[before]
ai, ax = sorder[after]
dist = abs(bi - ai)
if bx.seqid != ax.seqid or dist > gdist:
continue
start, end = range_minmax(((bx.start, bx.end), (ax.start, ax.end)))
start, end = max(start - bdist, 1), end + bdist
proxy = (bx.seqid, start, end)
for a, b in rows:
proxytrack[a] = proxy
tags = {}
if opts.bed:
bed = Bed(genomicblast, sorted=False)
key = lambda x: gene_name(x.accn.rsplit(".", 1)[0])
for query, bb in groupby(bed, key=key):
bb = list(bb)
if query not in proxytrack:
continue
proxy = proxytrack[query]
tag = "NS"
best_b = bb[0]
for b in bb:
hsp = (b.seqid, b.start, b.end)
if range_overlap(proxy, hsp):
tag = "S"
best_b = b
break
hsp = (best_b.seqid, best_b.start, best_b.end)
proxytrack[query] = hsp
tags[query] = tag
else:
blast = Blast(genomicblast)
for query, bb in blast.iter_hits():
bb = list(bb)
query = gene_name(query)
if query not in proxytrack:
continue
proxy = proxytrack[query]
tag = "NS"
best_b = bb[0]
for b in bb:
hsp = (b.subject, b.sstart, b.sstop)
if range_overlap(proxy, hsp):
tag = "S"
best_b = b
break
hsp = (best_b.subject, best_b.sstart, best_b.sstop)
proxytrack[query] = hsp
tags[query] = tag
for b in qbed:
accn = b.accn
target_region = genetrack[accn]
if accn in proxytrack:
target_region = region_str(proxytrack[accn])
if accn in tags:
ptag = "[{0}]".format(tags[accn])
else:
ptag = "[NF]"
target_region = ptag + target_region
print "\t".join((b.seqid, accn, target_region))
if emptyblast:
sh("rm -f {0}".format(genomicblast))
def rnaseq(args):
"""
%prog rnaseq blastfile ref.fasta
Evaluate de-novo RNA-seq assembly against a reference gene set (same or
closely related organism). Ideally blatfile needs to be supermap'd.
Following metric is used (Martin et al. 2010, Rnnotator paper):
Accuracy: % of contigs share >=95% identity with ref genome (TODO)
Completeness: % of ref genes covered by contigs to >=80% of their lengths
Contiguity: % of ref genes covered by a *single* contig >=80% of lengths
Chimer: % of contigs that contain two or more annotated genes >= 50bp
"""
from jcvi.algorithms.supermap import supermap
p = OptionParser(rnaseq.__doc__)
opts, args = p.parse_args(args)
if len(args) != 2:
sys.exit(p.print_help())
blastfile, reffasta = args
sizes = Sizes(reffasta).mapping
known_genes = len(sizes)
querysupermap = blastfile + ".query.supermap"
refsupermap = blastfile + ".ref.supermap"
if not op.exists(querysupermap):
supermap(blastfile, filter="query")
if not op.exists(refsupermap):
supermap(blastfile, filter="ref")
blast = Blast(querysupermap)
chimers = 0
goodctg80 = set()
goodctg50 = set()
for ctg, hits in blast.iter_hits():
bps = defaultdict(int)
for x in hits:
bps[x.subject] += abs(x.sstop - x.sstart) + 1
valid_hits = bps.items()
for vh, length in valid_hits:
rsize = sizes[vh]
ratio = length * 100. / rsize
if ratio >= 80:
goodctg80.add(ctg)
if ratio >= 50:
goodctg50.add(ctg)
# Chimer
if len(valid_hits) > 1:
chimers += 1
blast = Blast(refsupermap)
goodref80 = set()
goodref50 = set()
bps = defaultdict(int)
for x in blast.iter_line():
bps[x.subject] += abs(x.sstop - x.sstart) + 1
for vh, length in bps.items():
rsize = sizes[vh]
ratio = length * 100. / rsize
if ratio >= 80:
goodref80.add(vh)
if ratio >= 50:
goodref50.add(vh)
print >> sys.stderr, "Reference set: `{0}`, # of transcripts {1}".\
format(reffasta, known_genes)
print >> sys.stderr, "A total of {0} contigs map to 80% of a reference"\
" transcript".format(len(goodctg80))
print >> sys.stderr, "A total of {0} contigs map to 50% of a reference"\
" transcript".format(len(goodctg50))
print >> sys.stderr, "A total of {0} reference transcripts ({1:.1f}%) have 80% covered" \
.format(len(goodref80), len(goodref80) * 100. / known_genes)
