def test_nearest(self):
from cruzdb.models import Feature
f = Feature()
f.chrom = "chr1"
f.txStart = 10
f.txEnd = 61
#db = Genome('hg18', host="localhost", user="******")
db = self.dba
self.assert_(db.refGene.first() is not None)
self.assert_(db.refGene is not None)
res = db.knearest(db.refGene, f, k=2)
self.assert_(len(res) >= 2)
f = db.refGene.first()
key = (f.chrom, f.start, f.end, f.name)
for k in (2, 4, 6):
res = db.knearest("refGene", f, k=k)
assert len(res) >= k
self.assert_(
key in ((n.chrom, n.start, n.end, n.name) for n in res),
(res, f))
f = db.refGene.order_by(
db.refGene.txStart).filter(db.refGene.c.strand == "+").first()
assert f in db.upstream(db.refGene, f)
down = db.downstream(db.refGene, f, k=10)
self.assert_(len(down) >= 10)
self.assert_(all(d.start >= f.start for d in down))
def test_nearest(self):
from cruzdb.models import Feature
f = Feature()
f.chrom = "chr1"
f.txStart = 10
f.txEnd = 61
#db = Genome('hg18', host="localhost", user="******")
db = self.dba
self.assert_(db.refGene.first() is not None)
self.assert_(db.refGene is not None)
res = db.knearest(db.refGene, f, k=2)
self.assert_(len(res) >= 2)
f = db.refGene.first()
key = (f.chrom, f.start, f.end, f.name)
for k in (2, 4, 6):
res = db.knearest("refGene", f, k=k)
assert len(res) >= k
self.assert_(key in ((n.chrom, n.start, n.end, n.name) for n in res),
(res, f))
f = db.refGene.order_by(db.refGene.txStart).filter(db.refGene.c.strand == "+").first()
assert f in db.upstream(db.refGene, f)
down = db.downstream(db.refGene, f, k=10)
self.assert_(len(down) >= 10)
self.assert_(all(d.start >= f.start for d in down))
def setUp(self):
from cruzdb.models import Feature
self.f = Feature()
self.f.chrom = "chr1"
self.f.txStart = 10
self.f.txEnd = 61
self.f.cdsStart = 29
self.f.cdsEnd = 59
"""
+ exon
| coding-exon
_ UTR
- intron
10 20 26 29 34 39 47 52 59 61
++++++++++______+++|||||-----||||||||-----|||||||+++
# introns = [(20, 26), (34, 39), (47, 52)]
# coding introns = [(34, 39), (47, 52)]
"""
self.f.exonStarts = "10,26,39,52,"
self.f.exonEnds = "20,34,47,61,"
self.strand = self.f.strand = '+'
def annotate(g, fname, tables, feature_strand=False, in_memory=False,
header=None, out=sys.stdout, _chrom=None, parallel=False):
"""
annotate bed file in fname with tables.
distances are integers for distance. and intron/exon/utr5 etc for gene-pred
tables. if the annotation features have a strand, the distance reported is
negative if the annotation feature is upstream of the feature in question
if feature_strand is True, then the distance is negative if t
"""
close = False
if isinstance(out, basestring):
out = nopen(out, "w")
close = True
if parallel:
import multiprocessing
import signal
p = multiprocessing.Pool(initializer=lambda:
signal.signal(signal.SIGINT, signal.SIG_IGN))
chroms = _split_chroms(fname)
def write_result(fanno, written=[False]):
for i, d in enumerate(reader(fanno, header="ordered")):
if i == 0 and written[0] == False:
print >>out, "\t".join(d.keys())
written[0] = True
print >>out, "\t".join(x if x else "NA" for x in d.values())
os.unlink(fanno)
os.unlink(fanno.replace(".anno", ""))
for fchrom, (fout, fanno) in chroms:
p.apply_async(annotate, args=(g.db, fout.name, tables, feature_strand, True,
header, fanno, fchrom),
callback=write_result)
p.close()
p.join()
return out.name
if isinstance(g, basestring):
from . import Genome
g = Genome(g)
if in_memory:
from . intersecter import Intersecter
intersecters = [] # 1 per table.
for t in tables:
q = getattr(g, t) if isinstance(t, basestring) else t
if _chrom is not None:
q = q.filter_by(chrom=_chrom)
table_iter = q #page_query(q, g.session)
intersecters.append(Intersecter(table_iter))
elif isinstance(fname, basestring) and os.path.exists(fname) \
and sum(1 for _ in nopen(fname)) > 25000:
print >>sys.stderr, "annotating many intervals, may be faster using in_memory=True"
if header is None:
header = []
extra_header = []
for j, toks in enumerate(reader(fname, header=False)):
if j == 0 and not header:
if not (toks[1] + toks[2]).isdigit():
header = toks
if j == 0:
for t in tables:
annos = (getattr(g, t) if isinstance(t, basestring) else t).first().anno_cols
h = t if isinstance(t, basestring) else t._table.name if hasattr(t, "_table") else t.first()._table.name
extra_header += ["%s_%s" % (h, a) for a in annos]
if 0 != len(header):
if not header[0].startswith("#"):
header[0] = "#" + header[0]
print >>out, "\t".join(header + extra_header)
if header == toks: continue
if not isinstance(toks, ABase):
f = Feature()
f.chrom = toks[0]
f.txStart = int(toks[1])
f.txEnd = int(toks[2])
try:
f.strand = toks[header.index('strand')]
except ValueError:
pass
else:
f = toks
# for now, use the objects str to get the columns
# might want to use getattr on the original cols
toks = f.bed(*header).split("\t")
sep = "^*^"
for ti, tbl in enumerate(tables):
if in_memory:
objs = intersecters[ti].knearest(int(toks[1]), int(toks[2]), chrom=toks[0], k = 1)
else:
objs = g.knearest(tbl, toks[0], int(toks[1]), int(toks[2]), k=1)
if len(objs) == 0:
print >>out, "\t".join(toks + ["", "", ""])
continue
gp = hasattr(objs[0], "exonStarts")
names = [o.gene_name for o in objs]
if feature_strand:
strands = [-1 if f.is_upstream_of(o) else 1 for o in objs]
else:
strands = [-1 if o.is_upstream_of(f) else 1 for o in objs]
# dists can be a list of tuples where the 2nd item is something
# like 'island' or 'shore'
dists = [o.distance(f, features=gp) for o in objs]
pure_dists = [d[0] if isinstance(d, (tuple, list)) else d for d in dists]
# convert to negative if the feature is upstream of the query
for i, s in enumerate(strands):
if s == 1: continue
if isinstance(pure_dists[i], basestring): continue
pure_dists[i] *= -1
for i, (pd, d) in enumerate(zip(pure_dists, dists)):
if isinstance(d, tuple):
if len(d) > 1:
dists[i] = "%s%s%s" % (pd, sep, sep.join(d[1:]))
else:
dists[i] = pd
# keep uniqe name, dist combinations (occurs because of
# transcripts)
name_dists = set(["%s%s%s" % (n, sep, d) \
for (n, d) in zip(names, dists)])
name_dists = [nd.split(sep) for nd in name_dists]
# just take the first gene name if they are all the same
if len(set(nd[0] for nd in name_dists)) == 1:
toks.append(name_dists[0][0])
else:
toks.append(";".join(nd[0] for nd in name_dists))
# iterate over the feat type, dist cols
for i in range(1, len(name_dists[0])):
toks.append(";".join(nd[i] for nd in name_dists))
print >>out, "\t".join(toks)
if close:
out.close()
return out.name
def annotate(g,
fname,
tables,
feature_strand=False,
in_memory=False,
header=None,
out=sys.stdout,
_chrom=None,
parallel=False):
"""
annotate bed file in fname with tables.
distances are integers for distance. and intron/exon/utr5 etc for gene-pred
tables. if the annotation features have a strand, the distance reported is
negative if the annotation feature is upstream of the feature in question
if feature_strand is True, then the distance is negative if t
"""
close = False
if isinstance(out, basestring):
out = nopen(out, "w")
close = True
if parallel:
import multiprocessing
import signal
p = multiprocessing.Pool(
initializer=lambda: signal.signal(signal.SIGINT, signal.SIG_IGN))
chroms = _split_chroms(fname)
def write_result(fanno, written=[False]):
for i, d in enumerate(reader(fanno, header="ordered")):
if i == 0 and written[0] == False:
print >> out, "\t".join(d.keys())
written[0] = True
print >> out, "\t".join(d.values())
os.unlink(fanno)
os.unlink(fanno.replace(".anno", ""))
for fchrom, (fout, fanno) in chroms:
p.apply_async(annotate,
args=(g.db, fout.name, tables, feature_strand, True,
header, fanno, fchrom),
callback=write_result)
p.close()
p.join()
return out.name
if isinstance(g, basestring):
from . import Genome
g = Genome(g)
if in_memory:
from .intersecter import Intersecter
intersecters = [] # 1 per table.
for t in tables:
q = getattr(g, t) if isinstance(t, basestring) else t
if _chrom is not None:
q = q.filter_by(chrom=_chrom)
table_iter = q #page_query(q, g.session)
intersecters.append(Intersecter(table_iter))
elif isinstance(fname, basestring) and os.path.exists(fname) \
and sum(1 for _ in nopen(fname)) > 25000:
print >> sys.stderr, "annotating many intervals, may be faster using in_memory=True"
if header is None:
header = []
extra_header = []
for j, toks in enumerate(reader(fname, header=False)):
if j == 0 and not header:
if not (toks[1] + toks[2]).isdigit():
header = toks
if j == 0:
for t in tables:
annos = (getattr(g, t)
if isinstance(t, basestring) else t).first().anno_cols
h = t if isinstance(t,
basestring) else t._table.name if hasattr(
t, "_table") else t.first()._table.name
extra_header += ["%s_%s" % (h, a) for a in annos]
if 0 != len(header):
if not header[0].startswith("#"):
header[0] = "#" + header[0]
print >> out, "\t".join(header + extra_header)
if header == toks: continue
if not isinstance(toks, ABase):
f = Feature()
f.chrom = toks[0]
f.txStart = int(toks[1])
f.txEnd = int(toks[2])
try:
f.strand = toks[header.index('strand')]
except ValueError:
pass
else:
f = toks
# for now, use the objects str to get the columns
# might want to use getattr on the original cols
toks = f.bed(*header).split("\t")
sep = "^*^"
for ti, tbl in enumerate(tables):
if in_memory:
objs = intersecters[ti].knearest(int(toks[1]),
int(toks[2]),
chrom=toks[0],
k=1)
else:
objs = g.knearest(tbl,
toks[0],
int(toks[1]),
int(toks[2]),
k=1)
if len(objs) == 0:
print >> out, "\t".join(toks + ["", "", ""])
continue
gp = hasattr(objs[0], "exonStarts")
names = [o.gene_name for o in objs]
if feature_strand:
strands = [-1 if f.is_upstream_of(o) else 1 for o in objs]
else:
strands = [-1 if o.is_upstream_of(f) else 1 for o in objs]
# dists can be a list of tuples where the 2nd item is something
# like 'island' or 'shore'
dists = [o.distance(f, features=gp) for o in objs]
pure_dists = [
d[0] if isinstance(d, (tuple, list)) else d for d in dists
]
# convert to negative if the feature is upstream of the query
for i, s in enumerate(strands):
if s == 1: continue
if isinstance(pure_dists[i], basestring): continue
pure_dists[i] *= -1
for i, (pd, d) in enumerate(zip(pure_dists, dists)):
if isinstance(d, tuple):
if len(d) > 1:
dists[i] = "%s%s%s" % (pd, sep, sep.join(d[1:]))
else:
dists[i] = pd
# keep uniqe name, dist combinations (occurs because of
# transcripts)
name_dists = set(["%s%s%s" % (n, sep, d) \
for (n, d) in zip(names, dists)])
name_dists = [nd.split(sep) for nd in name_dists]
for i in range(len(name_dists[0])
): # iterate over the dist, feature, name cols
toks.append(";".join(nd[i] for nd in name_dists))
print >> out, "\t".join(toks)
if close:
out.close()
return out.name
