def download_with_sra_prefetch(aspera_key, outdir, pickle_prefix, ena_id):
command = ['prefetch', '', ena_id]
if aspera_key is not None:
_, ascp, _ = utils.run_command_popen_communicate(['which', 'ascp'], False, None, False)
command[1] = '-a {ascp}|{aspera_key}'.format(ascp=ascp.splitlines()[0], aspera_key=aspera_key)
run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, 3600, True)
if run_successfully:
_, prefetch_outdir, _ = utils.run_command_popen_communicate(['echo', '$HOME/ncbi/public/sra'], True, None,
False)
try:
os.rename(os.path.join(prefetch_outdir.splitlines()[0], ena_id + '.sra'),
os.path.join(outdir, ena_id + '.sra'))
except OSError as e:
print('Found the following error:'
'{}'.format(e))
from shutil import copy as shutil_copy
shutil_copy(os.path.join(prefetch_outdir.splitlines()[0], ena_id + '.sra'),
os.path.join(outdir, ena_id + '.sra'))
os.remove(os.path.join(prefetch_outdir.splitlines()[0], ena_id + '.sra'))
utils.save_variable_to_pickle(run_successfully, outdir, pickle_prefix + '.' + ena_id)
def gzip_files(file_2_compress, pickle_prefix, outdir):
if file_2_compress.endswith('.temp'):
out_file = os.path.splitext(file_2_compress)[0]
else:
out_file = file_2_compress
command = ['gzip', '--stdout', '--best', file_2_compress, '>', str(out_file + '.gz')]
run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, True, None, True)
if run_successfully:
os.remove(file_2_compress)
utils.save_variable_to_pickle(run_successfully, outdir,
str(pickle_prefix + '.' + os.path.basename(file_2_compress)))
def download_with_aspera(aspera_file_path, aspera_key, outdir, pickle_prefix, sra, ena_id):
command = ['ascp', '-QT', '-l', '300m', '', '-i', aspera_key, '', outdir]
if not sra:
command[4] = '-P33001'
command[7] = str('era-fasp@' + aspera_file_path)
pickle = pickle_prefix + '.' + aspera_file_path.rsplit('/', 1)[1]
else:
command[7] = '[email protected]:/sra/sra-instant/reads/ByRun/sra/{a}/{b}/{c}/{c}.sra'.format(
a=ena_id[:3], b=ena_id[:6], c=ena_id)
pickle = pickle_prefix + '.' + ena_id
run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, 3600, True)
utils.save_variable_to_pickle(run_successfully, outdir, pickle)
def download_with_curl(ftp_file_path, outdir, pickle_prefix, sra, ena_id):
command = ['curl', '--retry', '1', '', '-o', '']
if not sra:
command[3] = ftp_file_path
file_download = ftp_file_path.rsplit('/', 1)[1]
command[5] = os.path.join(outdir, file_download)
pickle = pickle_prefix + '.' + file_download
else:
command[3] = 'ftp://ftp-trace.ncbi.nih.gov/sra/sra-instant/reads/ByRun/sra/{a}/{b}/{c}/{c}.sra'.format(
a=ena_id[:3], b=ena_id[:6], c=ena_id)
command[5] = os.path.join(outdir, ena_id + '.sra')
pickle = pickle_prefix + '.' + ena_id
run_successfully, stdout, stderr = utils.run_command_popen_communicate(command, False, 3600, True)
utils.save_variable_to_pickle(run_successfully, outdir, pickle)