def call_file(filename):
out = []
try:
with get_fast5_file(filename, mode="r") as f5:
ftype = check_file_type(f5) # single-read/multi-read
for read in f5.get_reads():
read_id = read.read_id
run_id = read.run_id.decode('utf-8')
read_number = read.handle['Raw'].attrs[
'read_number'] if ftype == 'multi-read' else read.status.read_info[
0].read_number
start_time = read.handle['Raw'].attrs[
'start_time'] if ftype == 'multi-read' else read.status.read_info[
0].start_time
channel_number = read.handle[
read.global_key +
'channel_id'].attrs['channel_number'].decode('utf-8')
sampling_rate = read.handle[
read.global_key + 'channel_id'].attrs['sampling_rate']
exp_start_time = read.handle[
read.global_key +
'tracking_id'].attrs['exp_start_time'].decode('utf-8')
start_time = add_time_seconds(exp_start_time,
start_time / sampling_rate)
signal = read.get_raw_data()
signal = rescale_signal(signal)
basecall, qual = caller.call_raw_signal(signal)
out.append((read_id, run_id, read_number, channel_number,
start_time, basecall, qual))
except OSError:
return []
return out
def test_correct_type(self):
single_read_path = os.path.join(test_data, "single_reads", "read0.fast5")
single_read_id = Fast5File(single_read_path).get_read_id()
with get_fast5_file(single_read_path) as f5:
self.assertEqual(type(f5), Fast5File)
self.assertEqual(check_file_type(f5), SINGLE_READ)
self.assertEqual(len(f5.get_read_ids()), 1)
self.assertEqual(single_read_id, f5.get_read_ids()[0])
self.get_raw(f5)
multi_read_path = os.path.join(test_data, "multi_read", "batch_0.fast5")
with get_fast5_file(multi_read_path) as f5:
self.assertEqual(type(f5), MultiFast5File)
self.assertEqual(check_file_type(f5), MULTI_READ)
self.assertTrue(len(f5.get_read_ids()) >= 1)
self.get_raw(f5)
def check_file_type(myfile):
fobj = fast5_interface.get_fast5_file(os.path.join(root, name))
if fast5_interface.check_file_type(fobj) == "multi-read":
#convert file to single fast5
print("converting fast5 file****")
multi_to_single_fast5.convert_multi_to_single(os.path.join(root, name),
directory, "single")
def try_convert_read(input_file, output_handle):
with Fast5File(input_file, 'r') as single_f5:
file_type = check_file_type(single_f5)
if file_type != SINGLE_READ:
raise Fast5FileTypeError(
"Could not convert Single->Multi for file type '{}' with path '{}'"
"".format(file_type, input_file))
add_single_read_to_multi_fast5(output_handle, single_f5)
return os.path.basename(input_file)
def convert_fast5_type(directory):
#go through fast5 files and check if the files is multi or single fast5 file
#we need a single fast5 file
for root, dirs, files in os.walk(directory):
for name in files:
if name.endswith(".fast5"):
fobj = fast5_interface.get_fast5_file(os.path.join(root, name))
if fast5_interface.check_file_type(fobj) == "multi-read":
#convert file to single fast5
print("converting fast5 file****")
multi_to_single_fast5.convert_multi_to_single(
os.path.join(root, name), directory, "single")
def try_multi_to_single_conversion(input_file, output_folder, subfolder):
output_files = []
with MultiFast5File(input_file, 'r') as multi_f5:
file_type = check_file_type(multi_f5)
if file_type != MULTI_READ:
raise Fast5FileTypeError(
"Could not convert Multi->Single for file type '{}' with path '{}'"
"".format(file_type, input_file))
for read in multi_f5.get_reads():
try:
output_file = os.path.join(output_folder, subfolder,
"{}.fast5".format(read.read_id))
create_single_f5(output_file, read)
output_files.append(os.path.basename(output_file))
except Exception as e:
logger.error("{}\n\tFailed to copy read '{}' from {}"
"".format(str(e), read.read_id, input_file),
exc_info=exc_info)
return output_files