def __init__(self, fname=None, fileobj=None):
if fileobj:
self.bed = BedFile(fileobj=fileobj)
self.fname = '*fileobj*'
else:
self.bed = BedFile(fname)
self.fname = fname
Model.__init__(self)
def __init__(self, fname, nostrand=None):
self.regions = {} # store BED regions as keyed bins (chrom, bin)
self.fname = fname
if nostrand == 'nostrand':
self.nostrand = True
else:
self.nostrand = False
self.bed = BedFile(fname)
def bam_extract(inbam, outbam, bedfile, nostrand=False, quiet=False):
bed = BedFile(bedfile)
if not quiet:
eta = ETA(os.stat(bedfile).st_size, fileobj=bed)
else:
eta = None
passed = 0
for region in bed:
if eta:
eta.print_status(extra="extracted:%s" % (passed))
if not region.chrom in inbam.references:
continue
if not nostrand:
strand = region.strand
else:
strand = None
for read in bam_extract_reads(inbam, region.chrom, region.start,
region.end, strand):
outbam.write(read)
passed += 1
if not quiet:
eta.done()
sys.stderr.write("%s extracted\n" % (passed, ))
def __init__(self, fname, nostrand=None):
self.regions = {} # store BED regions as keyed bins (chrom, bin)
self.fname = fname
if nostrand == 'nostrand':
self.nostrand = True
else:
self.nostrand = False
self.bed = BedFile(fname)
def testBaseCallRegionStrand(self):
# confirm that BED strand *isn't* used
bam = MockBam(['test2'])
bam.add_read('foo1', 'atcgatcg', '........', 0, 0, cigar='8M')
bam.add_read('foo2', 'atcgatcg', 'AAAAAAAA', 0, 4, cigar='8M')
bam.add_read('foo3',
'accgatcg',
'########',
0,
4,
cigar='8M',
is_reverse=True)
bam.add_read('foo4', 'atcgactgatcg', '############', 0, 0, cigar='12M')
bed = BedFile(fileobj=StringIO.StringIO('''\
test2|4|7|foo|1|-
'''.replace('|', '\t')))
# remember: bed is 0-based, basecall is 1-based, so 4->7, corresponds to 5->8
out = StringIO.StringIO('')
ngsutils.bam.basecall.bam_basecall(bam,
os.path.join(
os.path.dirname(__file__),
'test.fa'),
showstrand=True,
regions=bed,
out=out)
valid = '''chrom|pos|ref|count|consensus call|minor call|ave mappings|entropy|A|C|G|T|N|Deletions|Gaps|Insertions|Inserts|+ strand %|A minor %|C minor %|G minor %|T minor %|N minor %|Deletion minor %|Insertion minor %
test2|5|A|4|A||1.0|1.87433193885|4|0|0|0|0|0|0|0||0.75|0.25|0.0|0.0|0.0|0.0|0.0|0.0
test2|6|T|4|T/C||1.0|2.94335833285|0|2|0|2|0|0|0|0||0.75|0.0|0.5|0.0|0.0|0.0|0.0|0.0
test2|7|C|4|C|T|1.0|3.45990045591|0|3|0|1|0|0|0|0||0.75|0.0|0.333333333333|0.0|0.0|0.0|0.0|0.0
test2|8|G|4|G||1.0|1.87433193885|0|0|4|0|0|0|0|0||0.75|0.0|0.0|0.25|0.0|0.0|0.0|0.0
'''.replace('|', '\t')
self.assertEqual(valid, out.getvalue())
bed = None
ref = None
stranded = True
include_name = False
last = None
for arg in sys.argv[1:]:
if last == '-min':
min_size = int(arg)
last = None
elif arg in ['-min']:
last = arg
elif arg == '-name':
include_name = True
elif arg == '-ns':
stranded = False
elif not bed and os.path.exists(arg):
bed = arg
elif not ref and os.path.exists(arg):
ref = arg
if not bed or not ref:
usage()
sys.exit(1)
bed_tofasta(BedFile(bed),
ref,
min_size=min_size,
stranded=stranded,
include_name=include_name)
#!/usr/bin/env python
'''
Tests for bedutils reduce
'''
import unittest
import StringIO
import ngsutils.bed.reduce
from ngsutils.bed import BedFile
from ngsutils.bam.t import _matches
bedtest = BedFile(fileobj=StringIO.StringIO('''\
chr1|100|150|foo1|10|+
chr1|140|200|foo2|10|+
chr1|500|550|foo3|10|+
chr1|560|600|foo4|10|-
chr1|600|700|foo5|10|+
'''.replace('|', '\t')))
class ReduceTest(unittest.TestCase):
def testReduce(self):
valids = '''\
chr1|100|200|foo1,foo2|20|+
chr1|500|550|foo3|10|+
chr1|560|700|foo4,foo5|20|+
'''.replace('|', '\t').split('\n')
out = StringIO.StringIO('')
ngsutils.bed.reduce.bed_reduce(bedtest,
if __name__ == '__main__':
qbed_fname = None
refbed_fname = None
maxdist = 100000
match = False
nostrand = False
last = None
for arg in sys.argv[1:]:
if last == '-max':
maxdist = int(arg)
last = None
elif arg in ['-max']:
last = arg
elif arg == '-match':
match = True
elif arg == '-nostrand':
nostrand = True
elif not qbed_fname and (os.path.exists(arg) or arg == '-'):
qbed_fname = arg
elif not refbed_fname and os.path.exists(arg):
refbed_fname = arg
if not qbed_fname or not refbed_fname:
usage()
qbed = BedFile(qbed_fname)
refbed = BedFile(refbed_fname)
find_nearest(qbed, refbed, maxdist, match, nostrand)
if last == '-qual':
min_qual = int(arg)
last = None
elif last == '-ref':
if os.path.exists(arg) and os.path.exists('%s.fai' % arg):
ref = arg
else:
print "Missing FASTA file or index: %s" % arg
usage()
last = None
elif last == '-count':
min_count = int(arg)
last = None
elif last == '-bed':
if os.path.exists(arg):
regions = BedFile(arg)
else:
print "BED file: %s not found!" % arg
usage()
last = None
elif last == '-mask':
mask = int(arg)
last = None
elif last == '-minorpct':
minorpct = float(arg)
last = None
elif last == '-profile':
profile = arg
last = None
elif arg == '-h':
usage()
def usage():
print __doc__
print """\
Usage: bedutils sizes bedfile
"""
sys.exit(1)
def bed_size(bed, out=sys.stdout):
for region in bed:
out.write('%s\n' % (region.end - region.start))
if __name__ == '__main__':
fname = None
for arg in sys.argv[1:]:
if arg == '-h':
usage()
if not fname and (os.path.exists(arg) or arg == '-'):
fname = arg
else:
print "Unknown option: %s" % arg
usage()
if not fname:
usage()
bed_size(BedFile(fname))
out.write('%s\n' % '\t'.join(cols))
if __name__ == '__main__':
fname1 = None
fname2 = None
stranded = True
for arg in sys.argv[1:]:
if arg == '-h':
usage()
elif arg == '-nostrand':
stranded = False
elif not fname1 and (arg == '-' or os.path.exists(arg)):
fname1 = arg
elif not fname2 and (arg == '-' or os.path.exists(arg)):
fname2 = arg
else:
print "Unknown option: %s" % arg
usage()
if not fname1 or fname2:
usage()
if fname1 == '-' and fname2 == '-':
usage("Both input files can't be from stdin!")
bed2 = BedFile(fname2)
bed_subtract(fname1, bed2, stranded)
bed2.close()
if __name__ == '__main__':
fname1 = None
fname2 = None
stranded = True
for arg in sys.argv[1:]:
if arg == '-h':
usage()
elif arg == '-nostrand':
stranded = False
elif not fname1 and (arg == '-' or os.path.exists(arg)):
fname1 = arg
elif not fname2 and (arg == '-' or os.path.exists(arg)):
fname2 = arg
else:
print "Unknown option: %s" % arg
usage()
if not fname1 or not fname2:
usage()
if fname1 == '-' and fname2 == '-':
usage("Both input files can't be from stdin!")
bed1 = BedStreamer(fname1)
bed2 = BedFile(fname2)
bed_subtract(bed1, bed2, stranded)
bed2.close()
if arg == '-h':
usage()
if last == '-name':
name = arg
last = None
elif last == '-score':
score = arg
last = None
elif last == '-strand':
strand = arg
last = None
elif last == '-rgb':
if not rgb_name:
rgb_name = arg
else:
rgb[rgb_name] = arg
rgb_name = None
last = None
elif arg in ['-name', '-score', '-strand', '-rgb']:
last = arg
elif not fname and (os.path.exists(arg) or arg == '-'):
fname = arg
else:
print "Unknown option: %s" % arg
usage()
if not fname:
usage()
bed_annotate(BedFile(fname), name=name, score=score, strand=strand, rgb=rgb)
#!/usr/bin/env python
'''
Tests for bedutils extend
'''
import unittest
import StringIO
from ngsutils.bed import BedFile
import ngsutils.bed.extend
testbed = BedFile(fileobj=StringIO.StringIO('''
chr1|10|90|foo|1|+
chr1|10|90|foo|1|-
chr1|100|150|foo|1|+
chr1|200|250|foo|1|-
'''.replace('|', '\t')))
class ExtendTest(unittest.TestCase):
def testAbsolute(self):
valid = '''chr1|10|110|foo|1|+
chr1|0|90|foo|1|-
chr1|100|200|foo|1|+
chr1|150|250|foo|1|-
'''.replace('|', '\t')
out = StringIO.StringIO('')
ngsutils.bed.extend.bed_extend(testbed, 100, relative=False, out=out)
self.assertEqual(out.getvalue(), valid)
#!/usr/bin/env python
'''
Tests for bedutils tobedgraph
'''
import unittest
import ngsutils.bed.tobedgraph
import StringIO
from ngsutils.bed import BedFile
bedtest = BedFile(fileobj=StringIO.StringIO('''\
test1|10|20|foo1|10|+
test1|10|20|foo2|10|-
test1|15|25|foo3|10|+
test1|100|150|foo4|1|+
'''.replace('|', '\t')))
class BedGraphTest(unittest.TestCase):
def testBedGraph(self):
valid = '''\
test1|10|15|2
test1|15|20|3
test1|20|25|1
test1|100|150|1
'''.replace('|', '\t')
sio = StringIO.StringIO("")
ngsutils.bed.tobedgraph.bed_tobedgraph(bedtest, out=sio)
self.assertEqual(valid, sio.getvalue())
sio.close()
present += 1
out.write('\t%s\n' % present)
if __name__ == '__main__':
refname = None
bedfiles = []
stranded = True
last = None
for arg in sys.argv[1:]:
if arg == '-h':
usage()
elif arg == '-ns':
stranded = False
elif not refname and os.path.exists(arg):
refname = arg
elif os.path.exists(arg):
bedfiles.append(BedFile(arg))
else:
print "Bad argument: %s" % arg
usage()
sys.exit(1)
if not refname or not bedfiles:
usage()
sys.exit(1)
bed_refcount(BedFile(refname), bedfiles, stranded=stranded)
if __name__ == "__main__":
bed = None
strand = None
norm = None
last = None
for arg in sys.argv[1:]:
if arg == '-h':
usage()
if last == '-norm':
norm = float(arg)
last = None
elif arg in ['-norm']:
last = arg
elif arg == '-plus':
strand = '+'
elif arg == '-minus':
strand = '-'
elif not bed and os.path.exists(arg):
bed = arg
else:
print "Unknown option or missing index: %s" % arg
usage()
if not bed:
usage()
bed_tobedgraph(BedFile(bed), strand, norm)
#!/usr/bin/env python
'''
Tests for bedutils refcount
'''
import unittest
import StringIO
import ngsutils.bed.refcount
from ngsutils.bed import BedFile
bedtest1 = BedFile('test1',
fileobj=StringIO.StringIO('''\
chr1|100|150|foo1|10|+
chr1|140|200|foo2|10|+
chr1|500|550|foo3|10|+
chr1|560|600|foo4|10|-
chr1|600|700|foo5|10|+
'''.replace('|', '\t')))
bedtest2 = BedFile('test2',
fileobj=StringIO.StringIO('''\
chr1|90|120|bar1|10|+
chr1|140|200|bar2|10|+
chr1|510|520|bar3|10|+
chr1|500|660|bar4|10|-
chr1|520|570|bar5|10|+
'''.replace('|', '\t')))
bedtest3 = BedFile('test3',
fileobj=StringIO.StringIO('''\
def bed_reduce(target_bed, query_bed, stranded=True, exact=False, out=sys.stdout):
for qregion in query_bed:
for tregion in target_bed.fetch(qregion.chrom, qregion.start, qregion.end, qregion.strand if stranded else None):
if not exact or (qregion.start == tregion.start and qregion.end == tregion.end):
qregion.write(out)
break
if __name__ == '__main__':
fnames = []
stranded = True
exact = False
for arg in sys.argv[1:]:
if arg == '-h':
usage()
if arg == '-nostrand':
stranded = False
elif arg == '-exact':
exact = True
elif os.path.exists(arg):
fnames.append(arg)
else:
print "Unknown option: %s" % arg
usage()
if not fnames or len(fnames) < 2:
usage()
bed_reduce(BedFile(fnames[0]), BedStreamer(fnames[1]), stranded, exact)
class ExcludeBED(object):
def __init__(self, fname, nostrand=None):
self.regions = {} # store BED regions as keyed bins (chrom, bin)
self.fname = fname
if nostrand == 'nostrand':
self.nostrand = True
else:
self.nostrand = False
self.bed = BedFile(fname)
# with open(fname) as f:
# for line in f:
# if not line:
# continue
# if line[0] == '#':
# continue
# cols = line.strip().split('\t')
# chrom = cols[0]
# start = int(cols[1])
# end = int(cols[2])
# if self.nostrand:
# strand = '?'
# else:
# strand = cols[5]
# startbin = start / 100000
# endbin = end / 100000
# for bin in xrange(startbin, endbin + 1):
# if not (chrom, bin) in self.regions:
# self.regions[(chrom, bin)] = []
# self.regions[(chrom, bin)].append((start, end, strand))
def filter(self, bam, read):
if not read.is_unmapped:
if self.nostrand:
strand = None
elif read.is_reverse:
strand = '-'
else:
strand = '+'
for region in self.bed.fetch(bam.getrname(read.tid), read.pos,
read.aend, strand):
# region found, exclude read
return False
return True
# bin = read.pos / 100000
# ref = bam.getrname(read.tid)
# if not (ref, bin) in self.regions:
# return True
# for start, end, strand in self.regions[(ref, bin)]:
# if not self.nostrand:
# if strand == '+' and read.is_reverse:
# continue
# if strand == '-' and not read.is_reverse:
# continue
# if start <= read.pos <= end:
# return False
# if start <= read.aend <= end:
# return False
# return True
def __repr__(self):
return 'Excluding from BED: %s%s' % (self.fname, ' nostrand'
if self.nostrand else '')
def close(self):
pass
''')
sys.exit(1)
if __name__ == '__main__':
bam_fname = None
bed_fname = None
maxdist = 100000
last = None
for arg in sys.argv[1:]:
if last == '-max':
maxdist = int(arg)
last = None
elif arg in ['-max']:
last = arg
elif not bam_fname and os.path.exists(arg):
bam_fname = arg
elif not bed_fname and os.path.exists(arg):
bed_fname = arg
if not bam_fname or not bed_fname:
usage()
bed = BedFile(bed_fname)
bam = ngsutils.bam.bam_open(bam_fname)
find_nearest(bam, bed, maxdist)
#!/usr/bin/env python
'''
Tests for bedutils tofasta
'''
import os
import unittest
import ngsutils.bed.tofasta
import StringIO
from ngsutils.bed import BedFile
bedtest = BedFile(fileobj=StringIO.StringIO('''\
test1|0|10|foo1|10|+
test1|10|20|foo2|10|-
test1|0|5|foo1|10|+
test3|0|50|foo1|10|+
'''.replace('|', '\t')))
fasta = os.path.join(os.path.dirname(__file__), 'test.fa')
class FASTATest(unittest.TestCase):
def testBedFASTA(self):
valid = '''\
>test1:0-10
aaaaaaaaaa
>test1:10-20
cccccccccc
'''
class ExcludeBED(object):
def __init__(self, fname, nostrand=None):
self.regions = {} # store BED regions as keyed bins (chrom, bin)
self.fname = fname
if nostrand == 'nostrand':
self.nostrand = True
else:
self.nostrand = False
self.bed = BedFile(fname)
# with open(fname) as f:
# for line in f:
# if not line:
# continue
# if line[0] == '#':
# continue
# cols = line.strip().split('\t')
# chrom = cols[0]
# start = int(cols[1])
# end = int(cols[2])
# if self.nostrand:
# strand = '?'
# else:
# strand = cols[5]
# startbin = start / 100000
# endbin = end / 100000
# for bin in xrange(startbin, endbin + 1):
# if not (chrom, bin) in self.regions:
# self.regions[(chrom, bin)] = []
# self.regions[(chrom, bin)].append((start, end, strand))
def filter(self, bam, read):
if not read.is_unmapped:
if self.nostrand:
strand = None
elif read.is_reverse:
strand = '-'
else:
strand = '+'
for region in self.bed.fetch(bam.getrname(read.tid), read.pos, read.aend, strand):
# region found, exclude read
return False
return True
# bin = read.pos / 100000
# ref = bam.getrname(read.tid)
# if not (ref, bin) in self.regions:
# return True
# for start, end, strand in self.regions[(ref, bin)]:
# if not self.nostrand:
# if strand == '+' and read.is_reverse:
# continue
# if strand == '-' and not read.is_reverse:
# continue
# if start <= read.pos <= end:
# return False
# if start <= read.aend <= end:
# return False
# return True
def __repr__(self):
return 'Excluding from BED: %s%s' % (self.fname, ' nostrand' if self.nostrand else '')
def close(self):
pass
def testBedFile(self):
fname = os.path.join(os.path.dirname(__file__), 'test.bed')
valid = ['chr1|100|150|foo|1|+',
'chr1|100|150|foo|1|-',
'chr1|200|250|foo|1|+',
'chr1|300|350|foo|1|-',
]
regions = ['%s|%s|%s|%s|%s|%s' % (x.chrom, x.start, x.end, x.name, x.score_int, x.strand) for x in BedFile(fname)]
self.assertTrue(_matches(valid, regions))
def bed_stats(infile, gtf_file=None, out=sys.stdout, quiet=False, names=False):
if not quiet:
sys.stderr.write('Calculating BED region stats...\n')
stats = BedStats(BedFile(infile), gtf_file, names=names)
stats.write(out)
def testBedFileObj(self):
valid = ['chr1|100|150|foo|1|+',
'chr1|100|150|foo|1|-',
]
instr = StringIO.StringIO('''
chr1|100|150|foo|1|+
chr1|100|150|foo|1|-
'''.replace('|', '\t'))
regions = ['%s|%s|%s|%s|%s|%s' % (x.chrom, x.start, x.end, x.name, x.score, x.strand) for x in BedFile(fileobj=instr)]
self.assertTrue(_matches(valid, regions))
def testBedRegion(self):
valid = ['chr1|100|150']
regions = ['%s|%s|%s' % (x.chrom, x.start, x.end) for x in BedFile(region="chr1:101-150")]
self.assertTrue(_matches(valid, regions))
Converts the "score" field to be an integer
"""
sys.exit(1)
def bed_clean(bed, out=sys.stdout):
for region in bed:
region.score = int(region.score)
if region.score > 1000:
region.score = 1000
out.write('%s\n' % region)
if __name__ == '__main__':
fname = None
for arg in sys.argv[1:]:
if arg == '-h':
usage()
if not fname and (os.path.exists(arg) or arg == '-'):
fname = arg
else:
print "Unknown option: %s" % arg
usage()
if not fname:
usage()
bed_clean(BedFile(fname))
