def testAnnotationIsPresent(self, track):
'''
PASS: File exists and is not empty
FAIL: File exists and is empty (no data except comments)
NA: File does not exist. This might indicate an error
or simply that the annotation has not been computed.
The value indicates the number of lines in the file.
'''
fn = PARAMS[track]
if not os.path.exists(fn):
return ('NA', 0)
nlines = IOTools.getNumLines(fn)
if nlines > 0:
return ('PASS', nlines)
else:
return ('FAIL', nlines)
def main(argv=None):
"""script main.
parses command line options in sys.argv, unless *argv* is given.
"""
if not argv:
argv = sys.argv
# setup command line parser
parser = E.OptionParser(version="%prog version: $Id$",
usage=globals()["__doc__"])
parser.add_option("-m",
"--min-overlap",
dest="min_overlap",
type="float",
help="minimum overlap [%default]")
parser.add_option("-a",
"--bam-file",
dest="filename_bam",
metavar="bam",
type="string",
help="bam-file to use (required) [%default]")
parser.add_option("-b",
"--bed-file",
dest="filename_bed",
metavar="bed",
type="string",
help="bed-file to use (required) [%default]")
parser.add_option("-s",
"--sort-bed",
dest="sort_bed",
action="store_true",
help="sort the bed file by chromosomal location before "
"processing. "
"[%default]")
parser.add_option(
"--assume-sorted",
dest="sort_bed",
action="store_false",
help="assume that the bed-file is sorted by chromosomal location. "
"[%default]")
parser.add_option(
"--split-intervals",
dest="split_intervals",
action="store_true",
help="treat split BAM intervals, for example spliced intervals, "
"as separate intervals. Note that a single alignment might be "
"counted several times as a result. "
"[%default]")
parser.set_defaults(
min_overlap=0.5,
filename_bam=None,
filename_bed=None,
sort_bed=True,
split_intervals=False,
)
# add common options (-h/--help, ...) and parse command line
(options, args) = E.start(parser, argv=argv)
filename_bam = options.filename_bam
filename_bed = options.filename_bed
if filename_bam is None and filename_bed is None:
if len(args) != 2:
raise ValueError(
"please supply a bam and a bed file or two bed-files.")
filename_bam, filename_bed = args
if filename_bed is None:
raise ValueError("please supply a bed file to compare to.")
if filename_bam is None:
raise ValueError("please supply a bam file to compare with.")
E.info("intersecting the two files")
min_overlap = options.min_overlap
options.stdout.write("category\talignments\n")
# get number of columns of reference bed file
for bed in Bed.iterator(IOTools.open_file(filename_bed)):
ncolumns_bed = bed.columns
break
E.info("assuming %s is bed%i format" % (filename_bed, ncolumns_bed))
if ncolumns_bed < 4:
raise ValueError("please supply a name attribute in the bed file")
# get information about
if filename_bam.endswith(".bam"):
format = "-abam"
samfile = pysam.AlignmentFile(filename_bam, "rb")
total = samfile.mapped
# latest bedtools uses bed12 format when bam is input
ncolumns_bam = 12
# count per read
sort_key = lambda x: x.name
else:
format = "-a"
total = IOTools.getNumLines(filename_bam)
# get bed format
ncolumns_bam = 0
for bed in Bed.iterator(IOTools.open_file(filename_bam)):
ncolumns_bam = bed.columns
break
if ncolumns_bam > 0:
E.info("assuming %s is bed%i fomat" % (filename_bam, ncolumns_bam))
if ncolumns_bam == 3:
# count per interval
sort_key = lambda x: (x.contig, x.start, x.end)
else:
# count per interval category
sort_key = lambda x: x.name
# use fields for bam/bed file (regions to count with)
data_fields = [
"contig", "start", "end", "name", "score", "strand", "thickstart",
"thickend", "rgb", "blockcount", "blockstarts", "blockends"
][:ncolumns_bam]
# add fields for second bed (regions to count in)
data_fields.extend([
"contig2", "start2", "end2", "name2", "score2", "strand2",
"thickstart2", "thickend2", "rgb2", "blockcount2", "blockstarts2",
"blockends2"
][:ncolumns_bed])
# add bases overlap
data_fields.append("bases_overlap")
data = collections.namedtuple("data", data_fields)
options.stdout.write("total\t%i\n" % total)
if total == 0:
E.warn("no data in %s" % filename_bam)
return
# SNS: sorting optional, off by default
if options.sort_bed:
bedcmd = "<( gunzip < %s | sort -k1,1 -k2,2n)" % filename_bed
else:
bedcmd = filename_bed
if options.split_intervals:
split = "-split"
else:
split = ""
# IMS: newer versions of intersectBed have a very high memory
# requirement unless passed sorted bed files.
statement = """bedtools intersect %(format)s %(filename_bam)s
-b %(bedcmd)s
%(split)s
-sorted -bed -wo -f %(min_overlap)f""" % locals()
E.info("starting counting process: %s" % statement)
proc = E.run(statement, return_popen=True, stdout=subprocess.PIPE)
E.info("counting")
counts_per_alignment = collections.defaultdict(int)
take_columns = len(data._fields)
def iterate(infile):
for line in infile:
if not line.strip():
continue
yield data._make(line[:-1].split()[:take_columns])
for read, overlaps in itertools.groupby(iterate(
IOTools.force_str(proc.stdout)),
key=sort_key):
annotations = [x.name2 for x in overlaps]
for anno in annotations:
counts_per_alignment[anno] += 1
for key, counts in sorted(counts_per_alignment.items()):
options.stdout.write("%s\t%i\n" % (key, counts))
# write footer and output benchmark information.
E.stop()