def __do_pair_align_and_comparison(self,
log_level=logging.WARNING,
pairing_batch_size=None,
seq_list_len=None,
fastq_subfmt="fastq-sanger"):
u.build_index(self.refseq_fn)
#--
start = time.time()
bwa_res = u.run_bwa_sampe(self.refseq_fn, self.read_fn, self.mate_fn)
print 'BWA_SAMPE: %f secs' % (time.time() - start)
#--
start = time.time()
py_res = u.run_bwa_py_sampe(self.refseq_fn, self.read_fn, self.mate_fn,
log_level, pairing_batch_size,
seq_list_len, fastq_subfmt)
print 'PYBWA_SAMPE: %f secs' % (time.time() - start)
#--
self.assertEqual(len(bwa_res), len(py_res))
for o, n in it.izip(bwa_res, py_res):
for k in o.keys():
if k == 'name':
nn = n[k].rsplit(
"/", 1)[0] # we are dropping the '/[12]' extension
self.assertEqual(o[k], nn)
elif k == 'aux':
self.assertEqual(len(o[k]), len(n[k]))
for x, y in it.izip(o[k], n[k]):
self.assertEqual(x, y)
else:
self.assertEqual(o[k], n[k])
def __do_pair_align_and_comparison(self, log_level=logging.WARNING,
pairing_batch_size=None,
seq_list_len=None,
fastq_subfmt="fastq-sanger"):
u.build_index(self.refseq_fn)
#--
start = time.time()
bwa_res = u.run_bwa_sampe(self.refseq_fn, self.read_fn, self.mate_fn)
print 'BWA_SAMPE: %f secs' % (time.time() - start)
#--
start = time.time()
py_res = u.run_bwa_py_sampe(self.refseq_fn, self.read_fn, self.mate_fn,
log_level, pairing_batch_size, seq_list_len,
fastq_subfmt)
print 'PYBWA_SAMPE: %f secs' % (time.time() - start)
#--
self.assertEqual(len(bwa_res), len(py_res))
for o, n in it.izip(bwa_res, py_res):
for k in o.keys():
if k == 'name':
nn = n[k].rsplit("/", 1)[0] # we are dropping the '/[12]' extension
self.assertEqual(o[k], nn)
elif k == 'aux':
self.assertEqual(len(o[k]), len(n[k]))
for x,y in it.izip(o[k], n[k]):
self.assertEqual(x, y)
else:
self.assertEqual(o[k],n[k])
def __compare_step_by_step(self):
if not self.opt.no_index:
u.build_index(self.refseq_fn)
#--
bwa_res = u.run_bwa_sampe(self.refseq_fn, self.read_fn, self.mate_fn)
py_res = u.run_bwa_py_sampe(self.refseq_fn, self.read_fn, self.mate_fn,
log_level=self.opt.log_level,
pairing_batch_size=self.opt.pairing_batch_size,
seq_list_len=self.opt.seq_list_len,
fastq_subfmt=self.opt.fastq_subfmt)
f = open("errors.dump", 'w')
error_count = 0
try:
self.assertEqual(len(bwa_res), len(py_res))
except AssertionError:
error_count += 1
f.write("Error on result n. %d\n" % i)
f.write("NUMBER OF RESULT FIELDS: %d, %d\n" % (
len(bwa_res), len(py_res)))
for i, (o, n) in enumerate(it.izip(bwa_res, py_res)):
try:
for k in o.keys():
if k == 'name':
nn = n[k].rsplit("/", 1)[0] # we are dropping the '/[12]' extension
self.assertEqual(o[k], nn)
elif k == 'aux':
self.assertEqual(len(o[k]), len(n[k]))
for x,y in it.izip(o[k], n[k]):
self.assertEqual(x, y)
else:
self.assertEqual(o[k],n[k])
except AssertionError:
error_count += 1
f.write("Error on result n. %d\n" % i)
for k in o.keys():
f.write("%r %r %r %r\n" % (k, o[k], n[k], o[k]==n[k]))
f.write("\n")
print "error count: %d" % error_count
f.close()