def elprep_sfm ():
# set up directories for intermediate results
file_in = sys.argv[1]
file_out = sys.argv[2]
input = os.path.basename(file_in)
output_prefix, output_extension = os.path.splitext(input)
stamp = str(time.time())
split_dir = os.path.join(os.getcwd(), "temp-" + stamp + os.sep)
result_dir = os.path.join(os.getcwd(), "temp-processed-" + stamp + os.sep)
os.mkdir(split_dir)
os.mkdir(result_dir)
# split command
nr_of_threads_opt = elprep_io_wrapper.cmd_option("--nr-of-threads", sys.argv)
elprep_io_wrapper.cmd_wrap_input(["elprep", "split"], file_in, split_dir, ["--output-prefix", output_prefix, "--output-type", "sam"] + nr_of_threads_opt)
# run filter command for split files
for root, dirs, files in os.walk(split_dir):
for file in files:
ffile = os.path.join(root, file)
processed_file = os.path.join(result_dir, os.path.basename(file))
elprep_io_wrapper.cmd_wrap_io(["elprep"], ffile, processed_file, sys.argv[3:])
os.remove(ffile)
os.rmdir(split_dir)
# merge command
elprep_io_wrapper.cmd_wrap_output(["elprep", "merge"], result_dir, file_out, nr_of_threads_opt)
# remove directories for intermediate results
for root, dirs, files in os.walk(result_dir):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
os.rmdir(result_dir)
def elprep_sfm_gnupar ():
# set up directories for intermediate results
file_in = sys.argv[1]
file_out = sys.argv[2]
input = os.path.basename(file_in)
output_prefix, output_extension = os.path.splitext(input)
stamp = str(time.time())
split_dir = os.path.join(os.getcwd(), "temp-" + stamp + os.sep)
result_dir = os.path.join(os.getcwd(), "temp-processed-" + stamp + os.sep)
os.mkdir(split_dir)
os.mkdir(result_dir)
# split command
nr_of_threads_opt = elprep_io_wrapper.cmd_option('--nr-of-threads', sys.argv)
intermediate_files_opt = elprep_io_wrapper.cmd_option('--intermediate-files-output-type', sys.argv)
if intermediate_files_opt:
intermediate_files_output_type = intermediate_files_opt[1]
else:
intermediate_files_output_type = "sam"
elprep_io_wrapper.cmd_wrap_input(["elprep", "split"], file_in, split_dir, ["--output-prefix", output_prefix, "--output-type", intermediate_files_output_type] + nr_of_threads_opt)
spread_file = os.path.join(split_dir, output_prefix + "-spread." + intermediate_files_output_type)
splits_path = os.path.join(split_dir, "splits" + os.sep)
# gnu parallel command
nr_of_jobs_opt = elprep_io_wrapper.cmd_option('--nr-of-jobs', sys.argv)
read_group_string = elprep_io_wrapper.cmd_option('--replace-read-group', sys.argv)
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(sys.argv[3:], '--nr-of-jobs')
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(given_cmd_opts, '--intermediate-files-output-type')
cmd_opts = given_cmd_opts
if read_group_string:
cmd_opts = elprep_io_wrapper.remove_cmd_option(cmd_opts, '--replace-read-group')
cmd_opts = cmd_opts + ['--replace-read-group', '\"' + read_group_string[1] + '\"']
cmd_opts = cmd_opts + ['--split-file']
cmd_list = ["elprep"]
elprep_cmd = '\'' + reduce(append_cmd, cmd_list + ['{}', result_dir + '{/.}.' + intermediate_files_output_type ] + cmd_opts) + '\''
gnu_cmd = 'parallel --gnu -j ' + str(nr_of_jobs_opt[1]) + ' ' + elprep_cmd + ' ::: ' + splits_path + '*.' + intermediate_files_output_type
subprocess.check_call(gnu_cmd, shell=True)
# command for spread file
spread_out_file = os.path.join(result_dir, output_prefix + "-spread." + intermediate_files_output_type)
elprep_io_wrapper.cmd_wrap_io(["elprep"], spread_file, spread_out_file , given_cmd_opts)
# merge command
elprep_io_wrapper.cmd_wrap_output(["elprep", "merge"], result_dir, file_out, nr_of_threads_opt)
# remove directories for intermediate results
for root, dirs, files in os.walk(splits_path):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
os.rmdir(splits_path)
os.remove(spread_file)
os.rmdir(split_dir)
for root, dirs, files in os.walk(result_dir):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
os.rmdir(result_dir)
def elprep_sfm():
# set up directories for intermediate results
file_in = sys.argv[1]
file_out = sys.argv[2]
input = os.path.basename(file_in)
output_prefix, output_extension = os.path.splitext(input)
stamp = str(time.time())
split_dir = os.path.join(os.getcwd(), "temp-" + stamp + os.sep)
result_dir = os.path.join(os.getcwd(), "temp-processed-" + stamp + os.sep)
os.mkdir(split_dir)
os.mkdir(result_dir)
# split command
nr_of_threads_opt = elprep_io_wrapper.cmd_option("--nr-of-threads", sys.argv)
intermediate_files_opt = elprep_io_wrapper.cmd_option("--intermediate-files-output-type", sys.argv)
if intermediate_files_opt:
intermediate_files_output_type = intermediate_files_opt[1]
else:
intermediate_files_output_type = "sam"
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(sys.argv[3:], "--intermediate-files-output-type")
cmd_opts = given_cmd_opts
elprep_io_wrapper.cmd_wrap_input(
["elprep", "split"],
file_in,
split_dir,
["--output-prefix", output_prefix, "--output-type", intermediate_files_output_type] + nr_of_threads_opt,
)
spread_file = os.path.join(split_dir, output_prefix + "-spread." + intermediate_files_output_type)
splits_path = os.path.join(split_dir, "splits" + os.sep)
# run filter command for split files
for root, dirs, files in os.walk(splits_path):
for file in files:
ext = os.path.splitext(file)[1]
if ext == ".sam" or ext == ".bam" or ext == ".cram":
ffile = os.path.join(root, file)
processed_file = os.path.join(result_dir, os.path.basename(file))
elprep_io_wrapper.cmd_wrap_io(["elprep"], ffile, processed_file, cmd_opts + ["--split-file"])
os.remove(ffile)
os.rmdir(splits_path)
# command for spread file
spread_out_file = os.path.join(result_dir, output_prefix + "-spread." + intermediate_files_output_type)
elprep_io_wrapper.cmd_wrap_io(["elprep"], spread_file, spread_out_file, cmd_opts)
os.remove(spread_file)
os.rmdir(split_dir)
# merge command
elprep_io_wrapper.cmd_wrap_output(["elprep", "merge"], result_dir, file_out, nr_of_threads_opt)
# remove directories for intermediate results
for root, dirs, files in os.walk(result_dir):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
os.rmdir(result_dir)
def elprep_sfm (argv):
# set up directories for intermediate results
file_in = argv[1]
file_out = argv[2]
input = os.path.basename(file_in)
output_prefix, output_extension = os.path.splitext(input)
stamp = str(time.time())
split_dir = os.path.join(os.getcwd(), "temp-" + stamp + os.sep)
result_dir = os.path.join(os.getcwd(), "temp-processed-" + stamp + os.sep)
os.mkdir(split_dir)
os.mkdir(result_dir)
# split command
nr_of_threads_opt = elprep_io_wrapper.cmd_option("--nr-of-threads", argv)
intermediate_files_opt = elprep_io_wrapper.cmd_option("--intermediate-files-output-type", argv)
if intermediate_files_opt:
intermediate_files_output_type = intermediate_files_opt[1]
else:
intermediate_files_output_type = "sam"
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(argv[3:], "--intermediate-files-output-type")
cmd_opts = given_cmd_opts
elprep_io_wrapper.cmd_wrap_input(["elprep", "split"], file_in, split_dir, ["--output-prefix", output_prefix, "--output-type", intermediate_files_output_type] + nr_of_threads_opt)
spread_file = os.path.join(split_dir, output_prefix + "-spread." + intermediate_files_output_type)
splits_path = os.path.join(split_dir, "splits" + os.sep)
# run filter command for split files
for root, dirs, files in os.walk(splits_path):
for file in files:
ext = os.path.splitext(file)[1]
if (ext == ".sam" or ext == ".bam" or ext == ".cram"):
ffile = os.path.join(root, file)
processed_file = os.path.join(result_dir, os.path.basename(file))
elprep_io_wrapper.cmd_wrap_io(["elprep"], ffile, processed_file, cmd_opts + ["--split-file"])
os.remove(ffile)
os.rmdir(splits_path)
# command for spread file
spread_out_file = os.path.join(result_dir, output_prefix + "-spread." + intermediate_files_output_type)
elprep_io_wrapper.cmd_wrap_io(["elprep"], spread_file, spread_out_file , cmd_opts)
os.remove(spread_file)
os.rmdir(split_dir)
# merge command
elprep_io_wrapper.cmd_wrap_output(["elprep", "merge"], result_dir, file_out, nr_of_threads_opt)
# remove directories for intermediate results
for root, dirs, files in os.walk(result_dir):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
os.rmdir(result_dir)
def elprep_sfm_gnupar(argv):
# set up directories for intermediate results
file_in = argv[1]
file_out = argv[2]
stamp = str(time.time())
split_dir = os.path.join(os.getcwd(), "temp-" + stamp + os.sep)
result_dir = os.path.join(os.getcwd(), "temp-processed-" + stamp + os.sep)
os.mkdir(split_dir)
os.mkdir(result_dir)
nr_of_jobs_opt = elprep_io_wrapper.cmd_option('--nr-of-jobs',
argv) or ['--nr-of-jobs', 1]
nr_of_threads_opt = elprep_io_wrapper.cmd_option(
'--nr-of-threads', argv) or [
'--nr-of-threads',
str((multiprocessing.cpu_count() / int(nr_of_jobs_opt[1])) or 1)
]
# split command
nr_of_split_merge_threads_opt = [
'--nr-of-threads',
str(int(nr_of_jobs_opt[1]) * int(nr_of_threads_opt[1]))
]
if os.path.isdir(file_in):
output_prefix, output_extension = os.path.splitext(
os.path.basename(os.listdir(file_in)[0]))
else: # file
output_prefix, output_extension = os.path.splitext(
os.path.basename(file_in))
if output_extension == '': # e.g. /dev/stdin
output_extension = '.sam'
intermediate_files_opt = elprep_io_wrapper.cmd_option(
'--intermediate-files-output-type', argv)
if intermediate_files_opt:
intermediate_files_output_type = intermediate_files_opt[1]
else:
intermediate_files_output_type = output_extension[1:]
intermediate_files_op_opt = elprep_io_wrapper.cmd_option(
'--intermediate-files-output-prefix', argv)
if intermediate_files_op_opt:
output_prefix = intermediate_files_op_opt[1]
single_end_opt = elprep_io_wrapper.flg_option("--single-end", argv)
elprep_io_wrapper.cmd_wrap_input(["elprep", "split"], file_in, split_dir, [
"--output-prefix", output_prefix, "--output-type",
intermediate_files_output_type
] + nr_of_split_merge_threads_opt + single_end_opt)
# gnu parallel command
read_group_string = elprep_io_wrapper.cmd_option('--replace-read-group',
argv)
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(
argv[3:], '--nr-of-jobs')
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(
given_cmd_opts, '--intermediate-files-output-type')
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(
given_cmd_opts, '--intermediate-files-output-prefix')
if single_end_opt:
splits_path = split_dir
given_cmd_opts = elprep_io_wrapper.remove_flg_option(
given_cmd_opts, "--single-end")
else:
splits_path = os.path.join(split_dir, "splits" + os.sep)
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(
given_cmd_opts, '--nr-of-threads')
cmd_opts = given_cmd_opts + nr_of_threads_opt
if read_group_string:
cmd_opts = elprep_io_wrapper.remove_cmd_option(cmd_opts,
'--replace-read-group')
cmd_opts = cmd_opts + [
'--replace-read-group', '\"' + read_group_string[1] + '\"'
]
cmd_list = ["elprep filter"]
elprep_cmd = '\'' + reduce(
append_cmd, cmd_list +
['{}', result_dir + '{/.}.' + intermediate_files_output_type] +
cmd_opts) + '\''
gnu_cmd = 'parallel --gnu -j ' + str(
nr_of_jobs_opt[1]
) + ' ' + elprep_cmd + ' ::: ' + splits_path + '*.' + intermediate_files_output_type
ret = subprocess.check_call(gnu_cmd, shell=True)
if ret != 0: raise SystemExit, ret
# command for spread file
if not single_end_opt:
spread_file = os.path.join(
split_dir,
output_prefix + "-spread." + intermediate_files_output_type)
spread_out_file = os.path.join(
result_dir,
output_prefix + "-spread." + intermediate_files_output_type)
elprep_io_wrapper.cmd_wrap_io(["elprep", "filter"], spread_file,
spread_out_file, given_cmd_opts +
nr_of_split_merge_threads_opt)
os.remove(spread_file)
# merge command
elprep_io_wrapper.cmd_wrap_output(["elprep", "merge"], result_dir,
file_out, nr_of_split_merge_threads_opt +
single_end_opt)
# remove directories for intermediate results
for root, dirs, files in os.walk(splits_path):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
if not single_end_opt:
os.rmdir(splits_path)
os.rmdir(split_dir)
for root, dirs, files in os.walk(result_dir):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
os.rmdir(result_dir)
def elprep_sfm_gnupar(argv):
# set up directories for intermediate results
file_in = argv[1]
file_out = argv[2]
input = os.path.basename(file_in)
output_prefix, output_extension = os.path.splitext(input)
stamp = str(time.time())
split_dir = os.path.join(os.getcwd(), "temp-" + stamp + os.sep)
result_dir = os.path.join(os.getcwd(), "temp-processed-" + stamp + os.sep)
os.mkdir(split_dir)
os.mkdir(result_dir)
# split command
nr_of_threads_opt = elprep_io_wrapper.cmd_option('--nr-of-threads', argv)
intermediate_files_opt = elprep_io_wrapper.cmd_option(
'--intermediate-files-output-type', argv)
if intermediate_files_opt:
intermediate_files_output_type = intermediate_files_opt[1]
else:
intermediate_files_output_type = "sam"
elprep_io_wrapper.cmd_wrap_input(["elprep", "split"], file_in, split_dir, [
"--output-prefix", output_prefix, "--output-type",
intermediate_files_output_type
] + nr_of_threads_opt)
spread_file = os.path.join(
split_dir, output_prefix + "-spread." + intermediate_files_output_type)
splits_path = os.path.join(split_dir, "splits" + os.sep)
# gnu parallel command
nr_of_jobs_opt = elprep_io_wrapper.cmd_option('--nr-of-jobs', argv)
read_group_string = elprep_io_wrapper.cmd_option('--replace-read-group',
argv)
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(
argv[3:], '--nr-of-jobs')
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(
given_cmd_opts, '--intermediate-files-output-type')
cmd_opts = given_cmd_opts
if read_group_string:
cmd_opts = elprep_io_wrapper.remove_cmd_option(cmd_opts,
'--replace-read-group')
cmd_opts = cmd_opts + [
'--replace-read-group', '\"' + read_group_string[1] + '\"'
]
cmd_opts = cmd_opts + ['--split-file']
cmd_list = ["elprep"]
elprep_cmd = '\'' + reduce(
append_cmd, cmd_list +
['{}', result_dir + '{/.}.' + intermediate_files_output_type] +
cmd_opts) + '\''
gnu_cmd = 'parallel --gnu -j ' + str(
nr_of_jobs_opt[1]
) + ' ' + elprep_cmd + ' ::: ' + splits_path + '*.' + intermediate_files_output_type
ret = subprocess.check_call(gnu_cmd, shell=True)
if ret != 0: raise SystemExit, ret
# command for spread file
spread_out_file = os.path.join(
result_dir,
output_prefix + "-spread." + intermediate_files_output_type)
elprep_io_wrapper.cmd_wrap_io(["elprep"], spread_file, spread_out_file,
given_cmd_opts)
# merge command
elprep_io_wrapper.cmd_wrap_output(["elprep", "merge"], result_dir,
file_out, nr_of_threads_opt)
# remove directories for intermediate results
for root, dirs, files in os.walk(splits_path):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
os.rmdir(splits_path)
os.remove(spread_file)
os.rmdir(split_dir)
for root, dirs, files in os.walk(result_dir):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
os.rmdir(result_dir)
def elprep(): elprep_io_wrapper.cmd_wrap_io(["elprep"], sys.argv[1], sys.argv[2], sys.argv[3:])
def elprep_sfm(argv):
# set up directories for intermediate results
file_in = argv[1]
file_out = argv[2]
stamp = str(time.time())
split_dir = os.path.join(os.getcwd(), "temp-" + stamp + os.sep)
result_dir = os.path.join(os.getcwd(), "temp-processed-" + stamp + os.sep)
os.mkdir(split_dir)
os.mkdir(result_dir)
# split command
nr_of_threads_opt = elprep_io_wrapper.cmd_option(
"--nr-of-threads",
argv) or ["--nr-of-threads",
str(multiprocessing.cpu_count())]
if os.path.isdir(file_in):
output_prefix, output_extension = os.path.splitext(
os.path.basename(os.listdir(file_in)[0]))
else: # file
output_prefix, output_extension = os.path.splitext(
os.path.basename(file_in))
if output_extension == '': # e.g. /dev/stdin
output_extension = '.sam'
intermediate_files_opt = elprep_io_wrapper.cmd_option(
"--intermediate-files-output-type", argv)
if intermediate_files_opt:
intermediate_files_output_type = intermediate_files_opt[1]
else:
intermediate_files_output_type = output_extension[1:]
file_out_output_prefix, file_out_output_extension = os.path.splitext(
os.path.basename(file_out))
fasta_opt = []
if intermediate_files_output_type == "cram" or file_out_output_extension == ".cram":
fasta_t_opt = elprep_io_wrapper.cmd_option("--reference-t", argv)
fasta_T_opt = elprep_io_wrapper.cmd_option("--reference-T", argv)
if fasta_t_opt:
fasta_opt = fasta_t_opt
elif fasta_T_opt:
fasta_opt = fasta_T_opt
else:
if intermediate_files_output_type == "cram":
print(
"Intermediate files output type is .cram, so need to pass either --reference-t or --reference-T"
)
return
else:
print(
"Output file output type is .cram, so need to pass either --reference-t or --reference-T"
)
return
intermediate_files_op_opt = elprep_io_wrapper.cmd_option(
"--intermediate-files-output-prefix", argv)
if intermediate_files_op_opt:
output_prefix = intermediate_files_op_opt[1]
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(
argv[3:], "--intermediate-files-output-type")
given_cmd_opts = elprep_io_wrapper.remove_cmd_option(
given_cmd_opts, "--intermediate-files-output-prefix")
cmd_opts = given_cmd_opts
nr_of_threads_opt_given = elprep_io_wrapper.cmd_option(
"--nr-of-threads", cmd_opts)
if not nr_of_threads_opt_given:
cmd_opts = cmd_opts + nr_of_threads_opt # so we pass --nr-of-threads to the elprep filter command explicitly
single_end_opt = elprep_io_wrapper.flg_option("--single-end", argv)
elprep_io_wrapper.cmd_wrap_input(["elprep", "split"], file_in, split_dir, [
"--output-prefix", output_prefix, "--output-type",
intermediate_files_output_type
] + fasta_opt + nr_of_threads_opt + single_end_opt)
if single_end_opt:
splits_path = split_dir
cmd_opts = elprep_io_wrapper.remove_flg_option(cmd_opts,
"--single-end")
else:
splits_path = os.path.join(split_dir, "splits" + os.sep)
# run filter command for split files
for root, dirs, files in os.walk(splits_path):
for file in files:
ext = os.path.splitext(file)[1]
if (ext == ".sam" or ext == ".bam" or ext == ".cram"):
ffile = os.path.join(root, file)
processed_file = os.path.join(result_dir,
os.path.basename(file))
elprep_io_wrapper.cmd_wrap_io(["elprep", "filter"], ffile,
processed_file, cmd_opts)
os.remove(ffile)
# command for spread file
# commands for split files and spread file are the same, but the files are stored in different folders
# we keep them in seperate folders for backwards compatibility with elPrep v2.61
# in a future release this will be simplified, so all split files will be in the same folder
if not single_end_opt:
os.rmdir(splits_path)
spread_file = os.path.join(
split_dir,
output_prefix + "-spread." + intermediate_files_output_type)
spread_out_file = os.path.join(
result_dir,
output_prefix + "-spread." + intermediate_files_output_type)
elprep_io_wrapper.cmd_wrap_io(["elprep", "filter"], spread_file,
spread_out_file, cmd_opts)
os.remove(spread_file)
os.rmdir(split_dir)
# merge command
elprep_io_wrapper.cmd_wrap_output(["elprep", "merge"], result_dir,
file_out, fasta_opt + nr_of_threads_opt +
single_end_opt)
# remove directories for intermediate results
for root, dirs, files in os.walk(result_dir):
for file in files:
ffile = os.path.join(root, file)
os.remove(ffile)
os.rmdir(result_dir)
def elprep_im(argv):
elprep_io_wrapper.cmd_wrap_io(["elprep", "filter"], argv[1], argv[2],
argv[3:])