def compute_fragments_statistics(settings, limit=None):
'''
'''
sam_file = settings["sam_file"]
image_file = settings["image_file"]
lengths = defaultdict(int)
zeros = 0
if not limit:
print "Process full SAM file..."
else:
print "Process %s lines from SAM file..." % limit
for i, sam_obj in enumerate(sc_sam_reader(sam_file)):
print i, "\r",
if limit and i > limit:
break
if sam_obj.TLEN > 0:
lengths[sam_obj.TLEN] += 1
else:
zeros += 1
keys = lengths.keys()
keys.sort()
for k in keys:
print k, lengths[k], round(100.*lengths[k]/i, 3)
print "Write image to %s" % image_file
draw_distribution_plot(lengths, image_file)
print "Fragments of zero length: %s" % zeros
def main(settings, limit=None):
'''
'''
sam_file = settings["sam_file"]
results = defaultdict(int)
for i, sam_obj in enumerate(sc_sam_reader(sam_file)):
results[sam_obj.FLAG] += 1
print results, "\r",
print
print "results", results
def main(settings, limit=None):
'''
'''
sam_file = settings["sam_file"]
output_file = settings["output_file"]
unmapped = 0
with open(output_file, "w") as fh:
for i, sam_obj in enumerate(sc_sam_reader(sam_file)):
if sam_obj.RNAME == "*":
fh.write(sam_obj.as_sam())
unmapped += 1
print i, unmapped, "\r",
print
print "Unmapped fraction:", float(unmapped)/(i+1)