def check_general_fails(self, filename, good_count):
handle = open(filename, "rU")
tuples = QualityIO.FastqGeneralIterator(handle)
for i in range(good_count):
title, seq, qual = next(tuples) # Make sure no errors!
self.assertRaises(ValueError, next, tuples)
handle.close()
def read_fastq(fastq: str, pair: Optional[str] = None) -> Tuple[Read]:
"""Read in a FASTQ file, and optionally its pair
'fastq' the filename for the forward or only FASTQ file
'pair' an optional filename for the reverse FASTQ file"""
reads = dict() # type: Dict[str, Read]
with open(fastq, 'r') as ffile: # type: _io.TextIOWrapper
for read in QualityIO.FastqGeneralIterator(ffile): # type: Tuple[str]
read_id, seq, qual = read # type: str, str, str
reads[read_id] = Read(read_id=read_id, seq=seq, qual=qual)
if pair:
with open(pair, 'r') as rfile: # type: _io.TextIOWrapper
for read in QualityIO.FastqGeneralIterator(
rfile): # type: Tuple[str]
read_id, seq, qual = read # type: str, str, str
reads[read_id].add_reverse(seq=seq, qual=qual)
return tuple(reads.values())
def check_general_fails(self, filename, good_count):
tuples = QualityIO.FastqGeneralIterator(filename)
msg = "FastqGeneralIterator failed to detect error in %s" % filename
for i in range(good_count):
title, seq, qual = next(tuples) # Make sure no errors!
# Detect error in the next record:
with self.assertRaises(ValueError, msg=msg) as cm:
title, seq, qual = next(tuples)
def check_general_passes(self, filename, record_count):
tuples = QualityIO.FastqGeneralIterator(filename)
# This "raw" parser doesn't check the ASCII characters which means
# certain invalid FASTQ files will get parsed without errors.
msg = "FastqGeneralIterator failed to parse %s" % filename
count = 0
for title, seq, qual in tuples:
self.assertEqual(len(seq), len(qual), msg=msg)
count += 1
self.assertEqual(count, record_count, msg=msg)
def check_general_passes(self, filename, record_count):
handle = open(filename, "rU")
tuples = QualityIO.FastqGeneralIterator(handle)
#This "raw" parser doesn't check the ASCII characters which means
#certain invalid FASTQ files will get parsed without errors.
count = 0
for title, seq, qual in tuples:
self.assertEqual(len(seq), len(qual))
count += 1
self.assertEqual(count, record_count)
handle.close()
def Main():
args = ParseArg()
# enable_thread_profiling()
# yappi.set_clock_type("wall")
# yappi.start()
# global type
# type = "fastq"
# if args.fastq:
# type="fastq"
# elif args.fasta:
# type="fasta"
global Range
Range = args.range
global Trim
Trim = args.trim
global max_score
max_score = args.max_score
name1 = args.input1.split('/')[-1]
name2 = args.input2.split('/')[-1]
#----------- read barcode ----------
global barcodes
global barcode_len
barcodes = []
for i in open(args.barcode, 'r'):
i = i.strip()
barcodes.append(i)
barcode_len = len(i)
barcodes.append('unassign')
#-----------------------------------
records = QualityIO.FastqGeneralIterator(open(args.input1, "rU"))
records2 = QualityIO.FastqGeneralIterator(open(args.input2, "rU"))
Files = MainIO(records, records2, name1, name2)
print "start to assign sequence to different barcodes..."
print "----------"
num_thread = args.parallel
# q=Queue(maxsize=10000)
# feeder = threading.Thread(target = push_stuff, args = (q, num_thread, records, records2))
# feeder.start()
workers = []
for i in range(num_thread):
worker = threading.Thread(target=do_stuff, args=(Files, ))
workers.append(worker)
#worker.setDaemon(True)
worker.start()
# print >>sys.stderr,"Finish reading records"
# q.join()
# feeder.join()
for i in range(num_thread):
workers[i].join()
