def __init__(self, bed, gtf_file=None, names=False):
self.regiontagger = None
if gtf_file:
self.regiontagger = RegionTagger(gtf_file)
self.total = 0
self.size = 0
self.lengths = Counts()
self.refs = {}
self.names = {}
for region in bed:
self.total += 1
self.size += (region.end - region.start)
self.lengths.add(region.end - region.start)
if names:
if not region.name in self.names:
self.names[region.name] = 1
else:
self.names[region.name] += 1
if not region.chrom in self.refs:
self.refs[region.chrom] = 0
self.refs[region.chrom] += 1
if self.regiontagger:
self.regiontagger.add_region(region.chrom, region.start,
region.end, region.strand)
def __init__(self, bamfile, gtf=None, region=None, delim=None, tags=[], show_all=False):
regiontagger = None
flag_counts = FlagCounts()
ref = None
start = None
end = None
if gtf:
regiontagger = RegionTagger(gtf, bamfile.references, only_first_fragment=True)
if region:
ref, startend = region.rsplit(':', 1)
if '-' in startend:
start, end = [int(x) for x in startend.split('-')]
start = start - 1
sys.stderr.write('Region: %s:%s-%s\n' % (ref, start + 1, end))
else:
start = int(startend) - 1
end = int(startend)
sys.stderr.write('Region: %s:%s\n' % (ref, start + 1))
total = 0
mapped = 0
unmapped = 0
tlen_counts = {}
names = set()
refs = {}
tagbins = {}
for tag in tags:
tagbins[tag] = FeatureBin(tag)
for rname in bamfile.references:
if delim:
refs[rname.split(delim)[0]] = 0
else:
refs[rname] = 0
# setup region or whole-file readers
def _foo1():
for read in bamfile.fetch(ref, start, end):
yield read
def _foo2():
for read in bam_iter(bamfile):
yield read
if region:
read_gen = _foo1
else:
read_gen = _foo2
has_ih = True
has_nh = True
try:
for read in read_gen():
if not show_all and read.is_paired and not read.is_read1:
# only operate on the first fragment
continue
try:
if has_ih and read.opt('IH') > 1:
if read.qname in names:
# reads only count once for this...
continue
names.add(read.qname)
except KeyError:
if not read.is_unmapped:
has_ih = False
#missing IH tag - ignore
pass
try:
if has_nh and read.opt('NH') > 1:
if read.qname in names:
# reads only count once for this...
continue
names.add(read.qname)
except KeyError:
if not read.is_unmapped:
has_nh = False
#missing NH tag - ignore
pass
flag_counts.add(read.flag)
total += 1
if read.is_unmapped:
unmapped += 1
continue
mapped += 1
if read.is_proper_pair and read.tid == read.mrnm:
# we don't care about reads that don't map to the same reference
# note: this doesn't work for RNA mapped to a reference genome...
# for RNA, you'd need to map to a transcript library (refseq) to get
# an accurate template length
#
# just skipping 'N' cigar values won't cut it either... since the pairs
# will likely silently span a gap.
if read.is_reverse:
k = -read.tlen
else:
k = read.tlen
if not k in tlen_counts:
tlen_counts[k] = 1
else:
tlen_counts[k] += 1
if delim:
refs[bamfile.getrname(read.rname).split(delim)[0]] += 1
else:
refs[bamfile.getrname(read.rname)] += 1
if regiontagger:
regiontagger.add_read(read, bamfile.getrname(read.rname))
for tag in tagbins:
tagbins[tag].add(read)
except KeyboardInterrupt:
sys.stderr.write('*** Interrupted - displaying stats up to this point! ***\n\n')
self.total = total
self.mapped = mapped
self.unmapped = unmapped
self.flag_counts = flag_counts
self.tagbins = tagbins
self.refs = refs
self.regiontagger = regiontagger
self.tlen_counts = tlen_counts