os.makedirs(options_storage.output_dir)
if options_storage.dataset_yaml_filename:
try:
dataset_data = pyyaml.load(file(options_storage.dataset_yaml_filename, "r"))
except pyyaml.YAMLError, exc:
support.error(
"exception caught while parsing YAML file (" + options_storage.dataset_yaml_filename + "):\n" + str(exc)
)
else:
dataset_data = support.correct_dataset(dataset_data)
options_storage.dataset_yaml_filename = os.path.join(options_storage.output_dir, "input_dataset.yaml")
pyyaml.dump(dataset_data, file(options_storage.dataset_yaml_filename, "w"))
support.check_dataset_reads(dataset_data, log)
if support.dataset_has_only_mate_pairs_libraries(dataset_data):
support.error(
"you should specify at least one paired-end or unpaired library (only mate-pairs libraries were found)!"
)
if options_storage.rectangles and (len(dataset_data) > 1):
support.error("rectangle graph algorithm for repeat resolution cannot work with multiple libraries!")
### FILLING cfg
cfg["common"] = empty_config()
cfg["dataset"] = empty_config()
if not options_storage.only_assembler:
cfg["error_correction"] = empty_config()
if not options_storage.only_error_correction:
cfg["assembly"] = empty_config()
# common
def fill_cfg(options_to_parse, log):
try:
options, not_options = getopt.gnu_getopt(options_to_parse,
options_storage.short_options,
options_storage.long_options)
except getopt.GetoptError:
_, exc, _ = sys.exc_info()
sys.stderr.write(str(exc) + "\n")
sys.stderr.flush()
options_storage.usage(spades_version)
sys.exit(1)
if not options:
options_storage.usage(spades_version)
sys.exit(1)
# all parameters are stored here
cfg = dict()
# dataset is stored here. We are prepared for up to MAX_LIBS_NUMBER paired-end libs and MAX_LIBS_NUMBER mate-pair libs
dataset_data = [{} for i in range(options_storage.MAX_LIBS_NUMBER * 2)]
options_storage.continue_mode = False
for opt, arg in options:
if opt == '-o':
options_storage.output_dir = arg
elif opt == "--tmp-dir":
options_storage.tmp_dir = arg
elif opt == "--reference":
options_storage.reference = support.check_file_existence(
arg, 'reference', log)
elif opt == "--dataset":
options_storage.dataset_yaml_filename = support.check_file_existence(
arg, 'dataset', log)
elif opt in options_storage.reads_options:
support.add_to_dataset(opt, arg, dataset_data)
elif opt == '-k':
options_storage.k_mers = list(map(int, arg.split(",")))
for k in options_storage.k_mers:
if k > 127:
support.error(
'wrong k value ' + str(k) +
': all k values should be less than 128', log)
if k % 2 == 0:
support.error(
'wrong k value ' + str(k) +
': all k values should be odd', log)
elif opt == "--sc":
options_storage.single_cell = True
elif opt == "--disable-gzip-output":
options_storage.disable_gzip_output = True
elif opt == "--only-error-correction":
if options_storage.only_assembler:
support.error(
'you cannot specify --only-error-correction and --only-assembler simultaneously'
)
options_storage.only_error_correction = True
elif opt == "--only-assembler":
if options_storage.only_error_correction:
support.error(
'you cannot specify --only-error-correction and --only-assembler simultaneously'
)
options_storage.only_assembler = True
elif opt == "--bh-heap-check":
options_storage.bh_heap_check = arg
elif opt == "--spades-heap-check":
options_storage.spades_heap_check = arg
elif opt == "--continue":
options_storage.continue_mode = True
elif opt == '-t' or opt == "--threads":
options_storage.threads = int(arg)
elif opt == '-m' or opt == "--memory":
options_storage.memory = int(arg)
elif opt == "--phred-offset":
if int(arg) in [33, 64]:
options_storage.qvoffset = int(arg)
else:
support.error(
'wrong PHRED quality offset value ' + str(arg) +
': should be either 33 or 64', log)
elif opt == '-i' or opt == "--iterations":
options_storage.iterations = int(arg)
elif opt == "--debug":
options_storage.developer_mode = True
elif opt == "--rectangles":
options_storage.rectangles = True
#corrector
elif opt == "--mismatch-correction":
options_storage.mismatch_corrector = True
elif opt == "--careful":
options_storage.mismatch_corrector = True
options_storage.careful = True
elif opt == '-h' or opt == "--help":
options_storage.usage(spades_version)
sys.exit(0)
elif opt == "--help-hidden":
options_storage.usage(spades_version, True)
sys.exit(0)
elif opt == "--test":
options_storage.set_test_options()
support.add_to_dataset(
'-1', os.path.join(spades_home,
"test_dataset/ecoli_1K_1.fq.gz"),
dataset_data)
support.add_to_dataset(
'-2', os.path.join(spades_home,
"test_dataset/ecoli_1K_2.fq.gz"),
dataset_data)
#break
else:
raise ValueError
if not options_storage.output_dir:
support.error(
"the output_dir is not set! It is a mandatory parameter (-o output_dir).",
log)
if not os.path.isdir(options_storage.output_dir):
if options_storage.continue_mode:
support.error("the output_dir should exist for --continue!", log)
os.makedirs(options_storage.output_dir)
if options_storage.continue_mode:
return None, None
if options_storage.dataset_yaml_filename:
try:
dataset_data = pyyaml.load(
open(options_storage.dataset_yaml_filename, 'r'))
except pyyaml.YAMLError:
_, exc, _ = sys.exc_info()
support.error('exception caught while parsing YAML file (' +
options_storage.dataset_yaml_filename + '):\n' +
str(exc))
dataset_data = support.relative2abs_paths(
dataset_data,
os.path.dirname(options_storage.dataset_yaml_filename))
else:
dataset_data = support.correct_dataset(dataset_data)
dataset_data = support.relative2abs_paths(dataset_data, os.getcwd())
options_storage.dataset_yaml_filename = os.path.join(
options_storage.output_dir, "input_dataset.yaml")
pyyaml.dump(dataset_data,
open(options_storage.dataset_yaml_filename, 'w'))
support.check_dataset_reads(dataset_data, options_storage.only_assembler,
log)
if support.dataset_has_only_mate_pairs_libraries(dataset_data):
support.error(
'you should specify at least one paired-end or unpaired library (only mate-pairs libraries were found)!'
)
if options_storage.rectangles and (len(dataset_data) > 1):
support.error(
'rectangle graph algorithm for repeat resolution cannot work with multiple libraries!'
)
### FILLING cfg
cfg["common"] = empty_config()
cfg["dataset"] = empty_config()
if not options_storage.only_assembler:
cfg["error_correction"] = empty_config()
if not options_storage.only_error_correction:
cfg["assembly"] = empty_config()
# common
cfg["common"].__dict__["output_dir"] = os.path.abspath(
options_storage.output_dir)
cfg["common"].__dict__["max_threads"] = options_storage.threads
cfg["common"].__dict__["max_memory"] = options_storage.memory
cfg["common"].__dict__["developer_mode"] = options_storage.developer_mode
# dataset section
cfg["dataset"].__dict__["single_cell"] = options_storage.single_cell
cfg["dataset"].__dict__["yaml_filename"] = os.path.abspath(
options_storage.dataset_yaml_filename)
if options_storage.developer_mode and options_storage.reference:
cfg["dataset"].__dict__["reference"] = options_storage.reference
# error correction
if (not options_storage.only_assembler) and (options_storage.iterations >
0):
cfg["error_correction"].__dict__["output_dir"] = os.path.join(
cfg["common"].output_dir, "corrected")
cfg["error_correction"].__dict__[
"max_iterations"] = options_storage.iterations
cfg["error_correction"].__dict__[
"gzip_output"] = not options_storage.disable_gzip_output
if options_storage.qvoffset:
cfg["error_correction"].__dict__[
"qvoffset"] = options_storage.qvoffset
if options_storage.bh_heap_check:
cfg["error_correction"].__dict__[
"heap_check"] = options_storage.bh_heap_check
if options_storage.tmp_dir:
cfg["error_correction"].__dict__[
"tmp_dir"] = options_storage.tmp_dir
else:
cfg["error_correction"].__dict__["tmp_dir"] = cfg[
"error_correction"].output_dir
cfg["error_correction"].tmp_dir = os.path.join(
os.path.abspath(cfg["error_correction"].tmp_dir), 'tmp')
# assembly
if not options_storage.only_error_correction:
if options_storage.k_mers:
cfg["assembly"].__dict__["iterative_K"] = options_storage.k_mers
else:
cfg["assembly"].__dict__[
"iterative_K"] = options_storage.k_mers_short
cfg["assembly"].__dict__["careful"] = options_storage.careful
if options_storage.spades_heap_check:
cfg["assembly"].__dict__[
"heap_check"] = options_storage.spades_heap_check
#corrector can work only if contigs exist (not only error correction)
if (not options_storage.only_error_correction
) and options_storage.mismatch_corrector:
cfg["mismatch_corrector"] = empty_config()
cfg["mismatch_corrector"].__dict__["skip-masked"] = ""
cfg["mismatch_corrector"].__dict__["bwa"] = os.path.join(
bin_home, "bwa-spades")
cfg["mismatch_corrector"].__dict__["threads"] = options_storage.threads
cfg["mismatch_corrector"].__dict__[
"output-dir"] = options_storage.output_dir
return cfg, dataset_data
def fill_cfg(options_to_parse, log):
try:
options, not_options = getopt.gnu_getopt(options_to_parse, options_storage.short_options, options_storage.long_options)
except getopt.GetoptError:
_, exc, _ = sys.exc_info()
sys.stderr.write(str(exc) + "\n")
sys.stderr.flush()
options_storage.usage(spades_version)
sys.exit(1)
if not options:
options_storage.usage(spades_version)
sys.exit(1)
# all parameters are stored here
cfg = dict()
# dataset is stored here. We are prepared for up to MAX_LIBS_NUMBER paired-end libs and MAX_LIBS_NUMBER mate-pair libs
dataset_data = [{} for i in range(options_storage.MAX_LIBS_NUMBER * 2)]
options_storage.continue_mode = False
for opt, arg in options:
if opt == '-o':
options_storage.output_dir = arg
elif opt == "--tmp-dir":
options_storage.tmp_dir = arg
elif opt == "--reference":
options_storage.reference = support.check_file_existence(arg, 'reference', log)
elif opt == "--dataset":
options_storage.dataset_yaml_filename = support.check_file_existence(arg, 'dataset', log)
elif opt in options_storage.reads_options:
support.add_to_dataset(opt, arg, dataset_data)
elif opt == '-k':
options_storage.k_mers = list(map(int, arg.split(",")))
for k in options_storage.k_mers:
if k > 127:
support.error('wrong k value ' + str(k) + ': all k values should be less than 128', log)
if k % 2 == 0:
support.error('wrong k value ' + str(k) + ': all k values should be odd', log)
elif opt == "--sc":
options_storage.single_cell = True
elif opt == "--disable-gzip-output":
options_storage.disable_gzip_output = True
elif opt == "--only-error-correction":
if options_storage.only_assembler:
support.error('you cannot specify --only-error-correction and --only-assembler simultaneously')
options_storage.only_error_correction = True
elif opt == "--only-assembler":
if options_storage.only_error_correction:
support.error('you cannot specify --only-error-correction and --only-assembler simultaneously')
options_storage.only_assembler = True
elif opt == "--bh-heap-check":
options_storage.bh_heap_check = arg
elif opt == "--spades-heap-check":
options_storage.spades_heap_check = arg
elif opt == "--continue":
options_storage.continue_mode = True
elif opt == '-t' or opt == "--threads":
options_storage.threads = int(arg)
elif opt == '-m' or opt == "--memory":
options_storage.memory = int(arg)
elif opt == "--phred-offset":
if int(arg) in [33, 64]:
options_storage.qvoffset = int(arg)
else:
support.error('wrong PHRED quality offset value ' + str(arg) + ': should be either 33 or 64', log)
elif opt == '-i' or opt == "--iterations":
options_storage.iterations = int(arg)
elif opt == "--debug":
options_storage.developer_mode = True
elif opt == "--rectangles":
options_storage.rectangles = True
#corrector
elif opt == "--mismatch-correction":
options_storage.mismatch_corrector = True
elif opt == "--careful":
options_storage.mismatch_corrector = True
options_storage.careful = True
elif opt == '-h' or opt == "--help":
options_storage.usage(spades_version)
sys.exit(0)
elif opt == "--help-hidden":
options_storage.usage(spades_version, True)
sys.exit(0)
elif opt == "--test":
options_storage.set_test_options()
support.add_to_dataset('-1', os.path.join(spades_home, "test_dataset/ecoli_1K_1.fq.gz"), dataset_data)
support.add_to_dataset('-2', os.path.join(spades_home, "test_dataset/ecoli_1K_2.fq.gz"), dataset_data)
#break
else:
raise ValueError
if not options_storage.output_dir:
support.error("the output_dir is not set! It is a mandatory parameter (-o output_dir).", log)
if not os.path.isdir(options_storage.output_dir):
if options_storage.continue_mode:
support.error("the output_dir should exist for --continue!", log)
os.makedirs(options_storage.output_dir)
if options_storage.continue_mode:
return None, None
if options_storage.dataset_yaml_filename:
try:
dataset_data = pyyaml.load(open(options_storage.dataset_yaml_filename, 'r'))
except pyyaml.YAMLError:
_, exc, _ = sys.exc_info()
support.error('exception caught while parsing YAML file (' + options_storage.dataset_yaml_filename + '):\n' + str(exc))
dataset_data = support.relative2abs_paths(dataset_data, os.path.dirname(options_storage.dataset_yaml_filename))
else:
dataset_data = support.correct_dataset(dataset_data)
dataset_data = support.relative2abs_paths(dataset_data, os.getcwd())
options_storage.dataset_yaml_filename = os.path.join(options_storage.output_dir, "input_dataset.yaml")
pyyaml.dump(dataset_data, open(options_storage.dataset_yaml_filename, 'w'))
support.check_dataset_reads(dataset_data, options_storage.only_assembler, log)
if support.dataset_has_only_mate_pairs_libraries(dataset_data):
support.error('you should specify at least one paired-end or unpaired library (only mate-pairs libraries were found)!')
if options_storage.rectangles and (len(dataset_data) > 1):
support.error('rectangle graph algorithm for repeat resolution cannot work with multiple libraries!')
### FILLING cfg
cfg["common"] = empty_config()
cfg["dataset"] = empty_config()
if not options_storage.only_assembler:
cfg["error_correction"] = empty_config()
if not options_storage.only_error_correction:
cfg["assembly"] = empty_config()
# common
cfg["common"].__dict__["output_dir"] = os.path.abspath(options_storage.output_dir)
cfg["common"].__dict__["max_threads"] = options_storage.threads
cfg["common"].__dict__["max_memory"] = options_storage.memory
cfg["common"].__dict__["developer_mode"] = options_storage.developer_mode
# dataset section
cfg["dataset"].__dict__["single_cell"] = options_storage.single_cell
cfg["dataset"].__dict__["yaml_filename"] = os.path.abspath(options_storage.dataset_yaml_filename)
if options_storage.developer_mode and options_storage.reference:
cfg["dataset"].__dict__["reference"] = options_storage.reference
# error correction
if (not options_storage.only_assembler) and (options_storage.iterations > 0):
cfg["error_correction"].__dict__["output_dir"] = os.path.join(cfg["common"].output_dir, "corrected")
cfg["error_correction"].__dict__["max_iterations"] = options_storage.iterations
cfg["error_correction"].__dict__["gzip_output"] = not options_storage.disable_gzip_output
if options_storage.qvoffset:
cfg["error_correction"].__dict__["qvoffset"] = options_storage.qvoffset
if options_storage.bh_heap_check:
cfg["error_correction"].__dict__["heap_check"] = options_storage.bh_heap_check
if options_storage.tmp_dir:
cfg["error_correction"].__dict__["tmp_dir"] = options_storage.tmp_dir
else:
cfg["error_correction"].__dict__["tmp_dir"] = cfg["error_correction"].output_dir
cfg["error_correction"].tmp_dir = os.path.join(os.path.abspath(cfg["error_correction"].tmp_dir), 'tmp')
# assembly
if not options_storage.only_error_correction:
if options_storage.k_mers:
cfg["assembly"].__dict__["iterative_K"] = options_storage.k_mers
else:
cfg["assembly"].__dict__["iterative_K"] = options_storage.k_mers_short
cfg["assembly"].__dict__["careful"] = options_storage.careful
if options_storage.spades_heap_check:
cfg["assembly"].__dict__["heap_check"] = options_storage.spades_heap_check
#corrector can work only if contigs exist (not only error correction)
if (not options_storage.only_error_correction) and options_storage.mismatch_corrector:
cfg["mismatch_corrector"] = empty_config()
cfg["mismatch_corrector"].__dict__["skip-masked"] = ""
cfg["mismatch_corrector"].__dict__["bwa"] = os.path.join(bin_home, "bwa-spades")
cfg["mismatch_corrector"].__dict__["threads"] = options_storage.threads
cfg["mismatch_corrector"].__dict__["output-dir"] = options_storage.output_dir
return cfg, dataset_data
