def extract_fast5(input_file_path, bin_h, mode='DNA'):
"""
Extract the signal and label from a single fast5 file
Args:
input_file_path: path of a fast5 file.
bin_h: handle of the binary file.
mode: The signal type dealed with. Default to 'DNA'.
"""
try:
(raw_data, raw_label, raw_start, raw_length) = labelop.get_label_raw(input_file_path, FLAGS.basecall_group,
FLAGS.basecall_subgroup)
except IOError:
fail_list.append(input_file_path)
return False
except:
fail_list.append(input_file_path)
return False
if mode=='rna':
print(type(raw_data))
raw_data = raw_data[::-1]
if FLAGS.normalization == 'mean':
raw_data = (raw_data - np.median(raw_data)) / np.float(np.std(raw_data))
elif FLAGS.normalization == 'median':
raw_data = (raw_data - np.median(raw_data)) / np.float(robust.mad(raw_data))
pre_start = raw_start[0]
pre_index = 0
for index, start in enumerate(raw_start):
if start - pre_start > FLAGS.length:
if index - 1 == pre_index:
# If a single segment is longer than the maximum singal length, skip it.
pre_start = start
pre_index = index
continue
event.append(np.pad(raw_data[pre_start:raw_start[index - 1]],
(0, FLAGS.length + pre_start - raw_start[index - 1]), mode='constant'))
event_length.append(int(raw_start[index - 1] - pre_start))
label_ind = raw_label['base'][pre_index:(index - 1)]
temp_label = [DNA_BASE[x.decode('UTF-8')] for x in label_ind]
label.append(
np.pad(temp_label, (0, FLAGS.length - index + 1 + pre_index), mode='constant', constant_values=-1))
label_length.append(index - 1 - pre_index)
pre_index = index - 1
pre_start = raw_start[index - 1]
if raw_start[index] - pre_start > FLAGS.length:
# Skip a single event segment longer than the required signal length
pre_index = index
pre_start = raw_start[index]
success_list.append(input_file_path)
while len(event) > FLAGS.batch:
for index in range(0, FLAGS.batch):
bin_h.write(struct.pack(format_string,
*[event_length[index]] + event[index].tolist() + [label_length[index]] + label[
index].tolist()))
del event[:FLAGS.batch]
del event_length[:FLAGS.batch]
del label[:FLAGS.batch]
del label_length[:FLAGS.batch]
return True
return False
def extract_file(input_file,output_file):
try:
(raw_data, raw_label, raw_start, raw_length) = labelop.get_label_raw(input_file,FLAGS.basecall_group,FLAGS.basecall_subgroup)
except IOError:
return False
except:
return False
f_signal = open(output_file+'.signal','w+')
f_label = open(output_file+'.label','w+')
f_signal.write(" ".join(str(val) for val in raw_data))
for index,start in enumerate(raw_start):
f_label.write("%d %d %c\n"%(start,start+raw_length[index],str(raw_label['base'][index])))
f_signal.close()
f_label.close()
return True
def extract_file(input_file):
try:
(raw_data, raw_label, raw_start,
raw_length) = labelop.get_label_raw(input_file, FLAGS.basecall_group,
FLAGS.basecall_subgroup)
except Exception as e:
print(str(e))
return False, (None, None)
raw_data_array = []
for index, start in enumerate(raw_start):
raw_data_array.append(
[start, start + raw_length[index],
str(raw_label['base'][index])])
if FLAGS.mode == 'rna':
raw_data = raw_data[::-1]
return True, (raw_data, np.array(raw_data_array, dtype='S8'))