def parse(self):
"""Parse file and store information in self."""
if isGzip(self.filepath):
with gzip.open(self.filepath, "rt") as FH:
self._parseFileHandle(FH)
else:
with open(self.filepath, "r") as FH:
self._parseFileHandle(FH)
def nbSeq(filepath):
"""
Return the number of sequences in file.
:param filepath: Path to the file.
:type filepath: str
:return: The number of sequences.
:rtype: int
"""
handler = open
handler_options = "r"
if isGzip(filepath):
handler = gzip.open
handler_options = "rt"
nb_lines = 0
with handler(filepath, handler_options) as reader:
for line in reader:
nb_lines += 1
return int(nb_lines / 4)
def nbSeq(filepath):
"""
Return the number of sequences in file.
:param filepath: Path to the file.
:type filepath: str
:return: The number of sequences.
:rtype: int
"""
nb_seq = 0
handler = open
handler_options = "r"
if isGzip(filepath):
handler = gzip.open
handler_options = "rt"
with handler(filepath, handler_options) as reader:
for line in reader:
if line.startswith(">"):
nb_seq += 1
return nb_seq
def __init__(self, filepath, mode="r"):
"""
Build and return an instance of FastqIO.
:param filepath: Path to the file.
:type filepath: str
:param mode: Mode to open the file ('r', 'w', 'a').
:type mode: str
:return: The new instance.
:rtype: FastqIO
"""
self.filepath = filepath
self.mode = mode
if (mode in ["w", "a"]
and filepath.endswith('.gz')) or (mode not in ["w", "a"]
and isGzip(filepath)):
self.file_handle = gzip.open(filepath, mode + "t")
else:
self.file_handle = open(filepath, mode)
self.current_line_nb = 1
def nbSeqAndNt(filepath):
"""
Return the number of sequences and nucleotids in file.
:param filepath: Path to the file.
:type filepath: str
:return: The number of sequences and the number of nucleotids.
:rtype: int, int
"""
nb_seq = 0
nb_nt = 0
handler = open
handler_options = "r"
if isGzip(filepath):
handler = gzip.open
handler_options = "rt"
with handler(filepath, handler_options) as reader:
for line in reader:
if line.startswith(">"):
nb_seq += 1
else:
nb_nt += len(line.rstrip())
return nb_seq, nb_nt
# Logger
logging.basicConfig(
format=
'%(asctime)s -- [%(filename)s][pid:%(process)d][%(levelname)s] -- %(message)s'
)
log = logging.getLogger(os.path.basename(__file__))
log.setLevel(logging.INFO)
log.info("Command: " + " ".join(sys.argv))
# Process
out_open_fct = open
out_mode = "w"
if args.output.endswith('.gz'):
out_open_fct = gzip.open
out_mode = "wt"
with out_open_fct(args.output, out_mode) as writer:
last_line = "\n"
for curr_in_file in args.inputs:
if not last_line.endswith("\n"):
writer.write("\n") # Start new line for a new file
in_open_fct = open
in_mode = "r"
if isGzip(curr_in_file):
in_open_fct = gzip.open
in_mode = "rt"
with in_open_fct(curr_in_file, in_mode) as reader:
for line in reader:
writer.write(line)
last_line = line
log.info("End of job")