def prepare_configs(src_config_dir, ds_args, log):
config_dir = os.path.join(ds_args.output_dir, "dipspades_configs")
copy_configs(src_config_dir, config_dir)
#log.info("dipSPAdes configs were copied to " + config_dir)
config_fname = os.path.join(config_dir, "config.info")
if not os.path.exists(config_fname):
support.check_file_existence(config_fname + ".template")
os.rename(config_fname + ".template", config_fname)
return os.path.abspath(config_fname)
def prepare_configs(src_config_dir, ds_args, log):
config_dir = os.path.join(ds_args.output_dir, "dipspades_configs")
copy_configs(src_config_dir, config_dir)
#log.info("dipSPAdes configs were copied to " + config_dir)
config_fname = os.path.join(config_dir, "config.info")
if not os.path.exists(config_fname):
support.check_file_existence(config_fname + ".template")
os.rename(config_fname + ".template", config_fname)
return os.path.abspath(config_fname)
def parse_arguments(argv, log):
try:
options, not_options = getopt.gnu_getopt(argv, DS_Args_List.short_options, DS_Args_List.long_options)
except getopt.GetoptError:
_, exc, _ = sys.exc_info()
sys.stderr.write(str(exc) + "\n")
sys.stderr.flush()
options_storage.usage("", dipspades=True)
sys.exit(1)
ds_args = DS_Args()
for opt, arg in options:
if opt == '-o':
ds_args.output_dir = os.path.abspath(arg)
elif opt == '--expect-gaps':
ds_args.allow_gaps = True
elif opt == '--expect-rearrangements':
ds_args.weak_align = True
elif opt == '--hap':
ds_args.haplocontigs_fnames.append(support.check_file_existence(arg, 'haplocontigs', log, dipspades=True))
elif opt == '-t' or opt == "--threads":
ds_args.max_threads = int(arg)
elif opt == '-m' or opt == "--memory":
ds_args.max_memory = int(arg)
elif opt == '--tmp-dir':
ds_args.tmp_dir = os.path.abspath(arg)
ds_args.haplocontigs = os.path.join(ds_args.output_dir, "haplocontigs")
if not ds_args.output_dir:
support.error("the output_dir is not set! It is a mandatory parameter (-o output_dir).", log, dipspades=True)
if not ds_args.haplocontigs_fnames:
support.error("cannot start dipSPAdes without at least one haplocontigs file!", log, dipspades=True)
if not ds_args.tmp_dir:
ds_args.tmp_dir = os.path.join(ds_args.output_dir, options_storage.TMP_DIR)
return ds_args
def parse_arguments(argv, log):
try:
options, not_options = getopt.gnu_getopt(argv,
DS_Args_List.short_options,
DS_Args_List.long_options)
except getopt.GetoptError:
_, exc, _ = sys.exc_info()
sys.stderr.write(str(exc) + "\n")
sys.stderr.flush()
options_storage.usage("", dipspades=True)
sys.exit(1)
ds_args = DS_Args()
for opt, arg in options:
if opt == '-o':
ds_args.output_dir = os.path.abspath(arg)
elif opt == '--expect-gaps':
ds_args.allow_gaps = True
elif opt == '--expect-rearrangements':
ds_args.weak_align = True
elif opt == '--hap':
ds_args.haplocontigs_fnames.append(
support.check_file_existence(arg,
'haplocontigs',
log,
dipspades=True))
elif opt == '-t' or opt == "--threads":
ds_args.max_threads = int(arg)
elif opt == '-m' or opt == "--memory":
ds_args.max_memory = int(arg)
elif opt == '--tmp-dir':
ds_args.tmp_dir = os.path.abspath(arg)
elif opt == '--dsdebug':
ds_args.dev_mode = True
elif opt == '--hap-assembly':
ds_args.haplotype_assembly = True
elif opt == '--dsK':
ds_args.k = int(arg)
ds_args.haplocontigs = os.path.join(ds_args.output_dir, "haplocontigs")
if not ds_args.output_dir:
support.error(
"the output_dir is not set! It is a mandatory parameter (-o output_dir).",
log,
dipspades=True)
if not ds_args.haplocontigs_fnames:
support.error(
"cannot start dipSPAdes without at least one haplocontigs file!",
log,
dipspades=True)
if not ds_args.tmp_dir:
ds_args.tmp_dir = os.path.join(ds_args.output_dir,
options_storage.TMP_DIR)
return ds_args
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)
if len(not_options) > 1:
for opt, arg in options:
if opt == "-k" and arg.strip().endswith(','):
support.error("Do not put spaces after commas in the list of k-mers sizes! Correct example: -k 21,33,55", log)
support.error("Please specify option (e.g. -1, -2, -s, etc) for the following paths: " + ", ".join(not_options[1:]) + "\n", log)
# all parameters are stored here
cfg = dict()
# dataset is stored here. We are prepared for up to MAX_LIBS_NUMBER for each type of short-reads libs
dataset_data = [{} for i in range(options_storage.MAX_LIBS_NUMBER * len(options_storage.SHORT_READS_TYPES.keys()))] # "[{}] * num" doesn't work here!
# for parsing options from "previous run command"
options_storage.continue_mode = False
options_storage.k_mers = None
for opt, arg in options:
if opt == '-o':
options_storage.output_dir = os.path.abspath(arg)
elif opt == "--tmp-dir":
options_storage.tmp_dir = os.path.abspath(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':
if arg == 'auto':
options_storage.k_mers = arg
else:
options_storage.k_mers = list(map(int, arg.split(",")))
for k in options_storage.k_mers:
if k < options_storage.MIN_K or k > options_storage.MAX_K:
support.error('wrong k value ' + str(k) + ': all k values should be between %d and %d' %
(options_storage.MIN_K, options_storage.MAX_K), 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 == "--iontorrent":
options_storage.iontorrent = True
elif opt == "--disable-gzip-output":
options_storage.disable_gzip_output = True
elif opt == "--disable-gzip-output:false":
options_storage.disable_gzip_output = False
elif opt == "--disable-rr":
options_storage.disable_rr = True
elif opt == "--disable-rr:false":
options_storage.disable_rr = False
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 == "--read-buffer-size":
options_storage.read_buffer_size = int(arg)
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 == "--restart-from":
if arg not in ['ec', 'as', 'mc'] and not arg.startswith('k'):
support.error("wrong value for --restart-from option: " + arg + " (only 'ec', 'as', 'k<int>', 'mc' are available)", log)
options_storage.continue_mode = True
options_storage.restart_from = arg
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 arg == 'auto':
options_storage.qvoffset = arg
elif 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 == "--debug:false":
options_storage.developer_mode = False
#corrector
elif opt == "--mismatch-correction":
options_storage.mismatch_corrector = True
elif opt == "--mismatch-correction:false":
options_storage.mismatch_corrector = False
elif opt == "--careful":
options_storage.mismatch_corrector = True
options_storage.careful = True
elif opt == "--careful:false":
options_storage.mismatch_corrector = False
options_storage.careful = False
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
elif opt == "--diploid":
options_storage.diploid_mode = True
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 and for --restart-from!", log)
os.makedirs(options_storage.output_dir)
if options_storage.restart_from:
if options_storage.continue_mode: # saving parameters specified with --restart-from
if not support.dataset_is_empty(dataset_data):
support.error("you cannot specify reads with --restart-from option!", log)
options_storage.save_restart_options(log)
else: # overriding previous run parameters
options_storage.load_restart_options()
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 not support.get_lib_ids_by_type(dataset_data, spades_logic.READS_TYPES_USED_IN_CONSTRUCTION):
support.error('you should specify at least one unpaired, paired-end, or high-quality mate-pairs library!')
options_storage.set_default_values()
### 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"] = options_storage.output_dir
cfg["common"].__dict__["tmp_dir"] = options_storage.tmp_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__["iontorrent"] = options_storage.iontorrent
cfg["dataset"].__dict__["yaml_filename"] = 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
cfg["error_correction"].__dict__["iontorrent"] = options_storage.iontorrent
# 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
cfg["assembly"].__dict__["disable_rr"] = options_storage.disable_rr
cfg["assembly"].__dict__["diploid_mode"] = options_storage.diploid_mode
if options_storage.spades_heap_check:
cfg["assembly"].__dict__["heap_check"] = options_storage.spades_heap_check
if options_storage.read_buffer_size:
cfg["assembly"].__dict__["read_buffer_size"] = options_storage.read_buffer_size
#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"] = None
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 check(self):
'''
Check correctness of parameters. Unless throws error.
'''
if self.multinom is None:
if self.model_func_file is None:
self.multinom = False
else:
self.multinom = True
if self.pop_labels is not None:
self.pop_labels = [x.strip() for x in self.pop_labels.split(',')]
if self.ns is not None:
self.ns = support.check_comma_sep_list(self.ns)
self.input_file = support.check_file_existence(self.input_file)
if self.resume_dir is not None:
self.resume_dir = support.check_dir_existence(self.resume_dir)
if self.resume_dir is not None and self.output_dir is None:
self.output_dir = support.ensure_dir_existence(
self.resume_dir + "_resumed", check_emptiness=True)
elif self.output_dir is None:
support.error("Parameter `Output directory` is required")
else:
self.output_dir = support.ensure_dir_existence(
self.output_dir, check_emptiness=True)
if self.input_file is None:
support.error(
"Parameter `Input file` is required")
if self.theta is None:
support.warning(
"`Theta0` is not specified. It would be 1.0.")
if self.gen_time is None:
support.warning(
"`Time for one generation` is not specified. Time will be in genetic units.")
self.input_data, self.ns, self.pop_labels = support.load_spectrum(
self.input_file, self.ns, self.pop_labels)
self.ns = np.array(self.ns)
self.number_of_populations = len(self.ns)
# Linked or unlinked data
if not self.linked_snp and self.boot_dir is not None:
support.warning(
"SNP's are marked as unlinked, so the directory with bootstrap will be ignored.")
elif self.linked_snp:
if self.boot_dir is not None:
self.boot_dir = support.check_dir_existence(self.boot_dir)
self.boots = gadma.Inference.load_bootstrap_data_from_dir(self.boot_dir, self.ns, self.pop_labels)
# Custom model
if self.model_func_file is not None:
self.model_func_file = support.check_file_existence(self.model_func_file)
file_with_model_func = imp.load_source('module', self.model_func_file)
try:
self.model_func = file_with_model_func.model_func
except:
support.error(
"File " + self.model_func_file + ' does not contain function named `model_func`.')
if self.model_func_file is not None:
if self.p_ids is not None:
self.p_ids = support.check_comma_sep_list(self.p_ids, is_int=False)
self.fracs = [float(x) for x in self.fracs.split(",")]
if len(self.fracs) != 3:
support.error(
"length of `Fractions` (Parameters of genetic algorithm) must be 3")
self.frac_of_old_models = self.fracs[0]
self.frac_of_mutated_models = self.fracs[1]
self.frac_of_crossed_models = self.fracs[2]
if self.moments_scenario and self.dadi_pts is not None:
support.warning(
"Moments doesn't use --pts argument, so it would be ignored")
if self.dadi_pts is None:
max_n = max(self.ns)
self.dadi_pts = [max_n, max_n + 10, max_n + 20]
else:
self.dadi_pts = support.check_comma_sep_list(self.dadi_pts)
self.put_default_structures()
self.final_check()
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
try:
options, datasets = getopt.gnu_getopt(sys.argv[1:], short_options,
long_options)
except getopt.GetoptError, err:
print str(err)
print ""
usage()
sys.exit(1)
for opt, arg in options:
if opt in ('-o', "--output-dir"):
output_dir = arg
make_latest_symlink = False
elif opt in ('-r', "--reference"):
support.check_file_existence(arg, "reference")
reference = arg
elif opt in ('-t', "--thread-num"):
thread_num = int(arg)
if thread_num < 1:
thread_num = 1
elif opt in ('-b', "--bin-size"):
if int(arg) > 0:
bin_size = int(arg)
elif opt in ('-k', "--kmer-size"):
if int(arg) > 0:
kmer = int(arg)
elif opt in ('-x', "--max-is"):
if int(arg) > 0:
max_is = int(arg)
elif opt in ('-s', "--skip-trimming"):
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)
if len(not_options) > 1:
for opt, arg in options:
if opt == "-k" and arg.strip().endswith(','):
support.error(
"Do not put spaces after commas in the list of k-mers sizes! Correct example: -k 21,33,55",
log)
support.error(
"Please specify option (e.g. -1, -2, -s, etc) for the following paths: "
+ ", ".join(not_options[1:]) + "\n", log)
# all parameters are stored here
cfg = dict()
# dataset is stored here. We are prepared for up to MAX_LIBS_NUMBER for each type of short-reads libs
dataset_data = [
{} for i in range(options_storage.MAX_LIBS_NUMBER *
len(options_storage.SHORT_READS_TYPES.keys()) +
len(options_storage.LONG_READS_TYPES))
] # "[{}]*num" doesn't work here!
# for parsing options from "previous run command"
options_storage.continue_mode = False
options_storage.k_mers = None
for opt, arg in options:
if opt == '-o':
options_storage.output_dir = os.path.abspath(arg)
elif opt == "--tmp-dir":
options_storage.tmp_dir = os.path.abspath(arg)
elif opt == "--configs-dir":
options_storage.configs_dir = support.check_dir_existence(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':
if arg == 'auto':
options_storage.k_mers = arg
else:
options_storage.k_mers = list(map(int, arg.split(",")))
for k in options_storage.k_mers:
if k < options_storage.MIN_K or k > options_storage.MAX_K:
support.error(
'wrong k value ' + str(k) +
': all k values should be between %d and %d' %
(options_storage.MIN_K, options_storage.MAX_K),
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 == "--iontorrent":
options_storage.iontorrent = True
elif opt == "--disable-gzip-output":
options_storage.disable_gzip_output = True
elif opt == "--disable-gzip-output:false":
options_storage.disable_gzip_output = False
elif opt == "--disable-rr":
options_storage.disable_rr = True
elif opt == "--disable-rr:false":
options_storage.disable_rr = False
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 == "--read-buffer-size":
options_storage.read_buffer_size = int(arg)
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 == "--restart-from":
if arg not in ['ec', 'as', 'mc'] and not arg.startswith('k'):
support.error(
"wrong value for --restart-from option: " + arg +
" (should be 'ec', 'as', 'k<int>', or 'mc'", log)
options_storage.continue_mode = True
options_storage.restart_from = arg
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 arg == 'auto':
options_storage.qvoffset = arg
elif arg in ['33', '64']:
options_storage.qvoffset = int(arg)
else:
support.error(
'wrong PHRED quality offset value: ' + arg +
' (should be either 33, 64, or \'auto\')', log)
elif opt == "--cov-cutoff":
if arg == 'auto' or arg == 'off':
options_storage.cov_cutoff = arg
elif support.is_float(arg) and float(arg) > 0.0:
options_storage.cov_cutoff = float(arg)
else:
support.error(
'wrong value for --cov-cutoff option: ' + arg +
' (should be a positive float number, or \'auto\', or \'off\')',
log)
elif opt == '-i' or opt == "--iterations":
options_storage.iterations = int(arg)
elif opt == "--debug":
options_storage.developer_mode = True
elif opt == "--debug:false":
options_storage.developer_mode = False
#corrector
elif opt == "--mismatch-correction":
options_storage.mismatch_corrector = True
elif opt == "--mismatch-correction:false":
options_storage.mismatch_corrector = False
elif opt == "--careful":
options_storage.mismatch_corrector = True
options_storage.careful = True
elif opt == "--careful:false":
options_storage.mismatch_corrector = False
options_storage.careful = False
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
elif opt == "--diploid":
options_storage.diploid_mode = True
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 and for --restart-from!",
log)
os.makedirs(options_storage.output_dir)
if options_storage.restart_from:
if options_storage.continue_mode: # saving parameters specified with --restart-from
if not support.dataset_is_empty(dataset_data):
support.error(
"you cannot specify reads with --restart-from option!",
log)
options_storage.save_restart_options(log)
else: # overriding previous run parameters
options_storage.load_restart_options()
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 not support.get_lib_ids_by_type(
dataset_data, spades_logic.READS_TYPES_USED_IN_CONSTRUCTION):
support.error(
'you should specify at least one unpaired, paired-end, or high-quality mate-pairs library!'
)
options_storage.set_default_values()
### 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"] = options_storage.output_dir
cfg["common"].__dict__["tmp_dir"] = options_storage.tmp_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__["iontorrent"] = options_storage.iontorrent
cfg["dataset"].__dict__[
"yaml_filename"] = 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
cfg["error_correction"].__dict__[
"iontorrent"] = options_storage.iontorrent
# 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
cfg["assembly"].__dict__["disable_rr"] = options_storage.disable_rr
cfg["assembly"].__dict__["diploid_mode"] = options_storage.diploid_mode
cfg["assembly"].__dict__["cov_cutoff"] = options_storage.cov_cutoff
if options_storage.spades_heap_check:
cfg["assembly"].__dict__[
"heap_check"] = options_storage.spades_heap_check
if options_storage.read_buffer_size:
cfg["assembly"].__dict__[
"read_buffer_size"] = options_storage.read_buffer_size
#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"] = None
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
log.addHandler(console)
check_binaries(bin_home, log)
# 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)]
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 = 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":
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
