def BamToBed12Iterator(handle,**kwargs):
'''
handle is an bam iterator
need references hash if handle is not filename.
'''
if type(handle)==type("string"):
handle=pysam.Samfile(handle,"rb");
for i in handle:
#print i #debug
if i.tid<0: continue
strand="+"
if i.is_reverse:
strand="-"
score=i.mapq
'''
test
'''
if kwargs.has_key("references"):
if isinstance(kwargs["references"],str):
chr=kwargs["references"]
else:
chr=kwargs["references"][i.tid];
else:
try:
chr=handle.references[i.tid];
except:
chr="chr"
if kwargs.has_key("strand"):
if kwargs["strand"]=="read1" or kwargs["strand"]=="firstMate":
read1=True
else:
read1=False
else:
read1=True
start=i.pos
end=i.aend
name=i.qname
cds_start=start
cds_end=start
itemRgb="0,0,0"
'''
debug
import sys
if i.cigar is None:
print >>sys.stderr,"why cigar is Nonetype?"
print >>sys.stderr,i
exit(0)
end of debug
'''
if i.cigar==None: continue # IGNORE THIS READS?
(block_starts,block_sizes)=Tools.cigar_to_coordinates(i.cigar);
if i.is_read1 and not read1:
strand=Tools.reverse_strand(strand)
elif i.is_read2 and read1:
strand=Tools.reverse_strand(strand)
bed=Bed12([chr,start,end,name,score,strand,cds_start,cds_end,itemRgb,len(block_sizes),block_sizes,block_starts])
yield bed
def BamToBed12Iterator(handle, **kwargs):
'''
handle is an bam iterator
need references hash if handle is not filename.
'''
if type(handle) == type("string"):
handle = pysam.Samfile(handle, "rb")
for i in handle:
#print i #debug
if i.tid < 0: continue
strand = "+"
if i.is_reverse:
strand = "-"
score = i.mapq
'''
test
'''
if kwargs.has_key("references"):
chr = kwargs["references"][i.tid]
else:
try:
chr = handle.references[i.tid]
except:
chr = "chr"
if kwargs.has_key("strand"):
if kwargs["strand"] == "read1" or kwargs["strand"] == "firstMate":
read1 = True
else:
read1 = False
else:
read1 = True
start = i.pos
end = i.aend
name = i.qname
cds_start = start
cds_end = start
itemRgb = "0,0,0"
'''
debug
import sys
if i.cigar is None:
print >>sys.stderr,"why cigar is Nonetype?"
print >>sys.stderr,i
exit(0)
end of debug
'''
if i.cigar == None: continue # IGNORE THIS READS?
(block_starts, block_sizes) = Tools.cigar_to_coordinates(i.cigar)
if i.is_read1 and not read1:
strand = Tools.reverse_strand(strand)
elif i.is_read2 and read1:
strand = Tools.reverse_strand(strand)
bed = Bed12([
chr, start, end, name, score, strand, cds_start, cds_end, itemRgb,
len(block_sizes), block_sizes, block_starts
])
yield bed
def query(self, x=None, method="pileup", **dict):
if type(x) == type("str"):
x = x.split(":")
chrom = x[0]
start = None
end = None
if len(x) > 1:
b = x[1].split("-")
if len(b) == 2:
start = int(b[0]) - 1
end = int(b[1])
elif isinstance(x, tuple):
chrom = x[CHROM]
start = x[CHROMSTART]
end = x[CHROMEND]
elif x is not None:
chrom = x.chr
start = x.start
end = x.stop
if method == "fetch":
for bamfile in self.bamfiles:
for read in bamfile.fetch(chrom, start, end):
if read.tid < 0:
continue
if read.mapq == 0:
continue
strand = "+"
if read.is_reverse:
strand = "-"
score = read.mapq
bed = Bed([bamfile.references[read.tid], read.pos, read.aend, read.qname, score, strand])
yield bed
elif method == "fetch12":
"""
test version
still test Tools.cigar_to_coordinates
"""
for bamfile in self.bamfiles:
for read in bamfile.fetch(chrom, start, end):
if read.tid < 0:
continue
if read.mapq == 0:
continue
chr = bamfile.references[read.tid]
strand = "+"
if read.is_reverse:
strand = "-"
score = read.mapq
start = read.pos
end = read.aend
name = read.qname
cds_start = start
cds_end = start
itemRgb = "0,0,0"
(block_starts, block_sizes) = Tools.cigar_to_coordinates(read.cigar)
bed = Bed12(
[
chr,
start,
end,
name,
score,
strand,
cds_start,
cds_end,
itemRgb,
len(block_sizes),
block_sizes,
block_starts,
]
)
yield bed
elif method == "bam1":
# fetch read from paired end with strand information
for bamfile in self.bamfiles:
strand = "read1"
if dict.has_key("strand"): # TODO: if bamfiles have different read1 or read2 ?
strand = dict["strand"]
for bed in TableIO.parse(
bamfile.fetch(chrom, start, end), "bam2bed12", references=chrom, strand=strand
):
yield bed
elif method == "paired_end":
for bamfile in self.bamfiles:
for fragment in TableIO.parse(bamfile.fetch(chrom, start, end), "bam2fragment", bam=bamfile):
yield fragment
elif method == "bam2": # yield bed12
for bamfile in self.bamfiles:
for fragment in TableIO.parse(bamfile.fetch(chrom, start, end), "bam2fragment", bam=bamfile):
if dict.has_key("strand"):
yield fragment.toBed12(chr=chrom, strand=dict["strand"])
else:
yield fragment.toBed12(chr=chrom)
elif method == "bam1tuple":
for bamfile in self.bamfiles:
strand = "read1"
if dict.has_key("strand"): # TODO: if bamfiles have different read1 or read2 ?
strand = dict["strand"]
for bed in TableIO.parse(
bamfile.fetch(chrom, start, end), "bam2bed12tuple", references=chrom, strand=strand
):
yield bed
elif method == "bam2tuple":
for bamfile in self.bamfiles:
for fragment in TableIO.parse(bamfile.fetch(chrom, start, end), "bam2fragment", bam=bamfile):
if dict.has_key("strand"):
yield fragment.toBed12Tuple(chr=chrom, strand=dict["strand"])
else:
yield fragment.toBed12Tuple(chr=chrom)
elif method == "bam2tuple_fast":
for bamfile in self.bamfiles:
for fragment in TableIO.parse(bamfile.fetch(chrom, start, end), "bam2fragment"):
if dict.has_key("strand"):
yield fragment.toBed12Tuple(chr=chrom, strand=dict["strand"])
else:
yield fragment.toBed12Tuple(chr=chrom)
elif method == "pileup":
s = [[0, 0, 0, 0] for row in range(end - start)]
for bamfile in self.bamfiles:
try:
A = bamfile.pileup(chrom, start, end)
except:
print >>sys.stderr, "Can't pile up", chrom, start, end
raise StopIteration
for pileupcolumn in A:
j = pileupcolumn.pos - start
if j < 0:
continue
if j > end - start:
break
for pileupread in pileupcolumn.pileups:
try:
if pileupread.is_del:
continue
if pileupread.indel != 0:
continue
nt = pileupread.alignment.seq[pileupread.qpos]
if BamI.hNtToNum.has_key(nt):
k = BamI.hNtToNum[nt]
s[j][k] += 1
except:
pass
for i in s:
yield i
elif method == "count":
s = 0
for bamfile in self.bamfiles:
s += bamfile.count(chrom, start, end)
yield s
elif method == "count_fragment":
s = 0
for bamfile in self.bamfiles:
for fragment in TableIO.parse(bamfile.fetch(chrom, start, end), "bam2fragment", bam=bamfile):
s += 1
yield s
elif method == "references":
for i in self.bamfiles[0].references:
yield i
elif method == "lengths":
for i in self.bamfiles[0].lengths:
yield i
def query(self,x,method='pileup'):
if method=='fetch':
for bamfile in self.bamfiles:
for read in bamfile.fetch(x.chr,x.start,x.stop):
if read.tid<0:continue
if read.mapq==0:continue
strand='+'
if read.is_reverse:
strand='-'
score=read.mapq
bed=Bed([bamfile.references[read.tid],read.pos,read.aend,read.qname,score,strand])
yield bed
elif method=='fetch12':
'''
test version
still test Tools.cigar_to_coordinates
'''
for bamfile in self.bamfiles:
for read in bamfile.fetch(x.chr,x.start,x.stop):
if read.tid<0:continue
if read.mapq==0:continue
chr=bamfile.references[read.tid]
strand='+'
if read.is_reverse:
strand='-'
score=read.mapq
start=read.pos
end=read.aend
name=read.qname
cds_start=start
cds_end=start
itemRgb="0,0,0"
(block_starts,block_sizes)=Tools.cigar_to_coordinates(read.cigar);
bed=Bed12([chr,start,end,name,score,strand,cds_start,cds_end,itemRgb,len(block_sizes),block_sizes,block_starts])
yield bed
elif method=="paired_end":
for bamfile in self.bamfiles:
for fragment in TableIO.parse(bamfile.fetch(x.chr,x.start,x.stop),"bam2fragment",bam=bamfile):
yield fragment
elif method=='pileup':
s=[[0,0,0,0] for row in range(x.stop-x.start)]
for bamfile in self.bamfiles:
try:
A=bamfile.pileup(x.chr,x.start,x.stop)
except:
print >>sys.stderr,"Can't pile up",x.chr,x.start,x.stop
raise StopIteration
for pileupcolumn in A:
j=pileupcolumn.pos-x.start
if j<0: continue
if j>x.stop-x.start: break
for pileupread in pileupcolumn.pileups:
try:
if pileupread.is_del: continue
if pileupread.indel!=0: continue
nt=pileupread.alignment.seq[pileupread.qpos]
if BamI.hNtToNum.has_key(nt):
k=BamI.hNtToNum[nt]
s[j][k]+=1
except:
pass
for i in s:
yield i