def check_arguments(args):
errors = []
errors += check_inputs(args, stdin_allowed=True)
errors += check_outputs(args, stdout_allowed=True)
errors += check_cpu_count(args)
errors += check_file_arg(args.seqids_path, none_allowed=True,
prefix='-S/--seqids_path')
errors += check_file_arg(args.descriptions_path, none_allowed=True,
prefix='-D/--descriptions_path')
errors += check_file_arg(args.names_path, none_allowed=True,
prefix='-N/--names_path')
errors += check_file_arg(args.mol_types_path, none_allowed=True,
prefix='-M/--mol_types_path')
errors += check_file_arg(args.taxids_path, none_allowed=True,
prefix='-T/--taxids_path')
if args.input_format != 'genbank':
values = [args.names, args.mol_types, args.taxids]
paths = [args.names_path, args.mol_types_path, args.taxids_path]
prefixes = ['-n/-N', '-m/-M', '-t/-T']
for value, path, prefix in zip(values, paths, prefixes):
if value or path is not None:
errors += ('{}: These options require input file(s) '
'in GenBank format.'.format(prefix))
if args.comparison is None:
args.comparison = COMPARISON
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
def check_arguments(args):
errors = []
# Insert argument verifications here
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
def print_missing_seqids(seqids_found):
not_found = []
for seqid in seqids_found:
if not seqids_found[seqid]:
not_found.append(seqid)
if not_found:
print_stderr('{:d} identifier(s) not found: {}'.format(
len(not_found), ', '.join(not_found)))
def import_annotations(annot_paths,
types,
levels,
obsolete,
main_ids,
add_ancestors,
ancestors,
min_level=MIN_LEVEL,
max_level=MAX_LEVEL):
bp, mf, cc, ec = {}, {}, {}, {}
bp_anc, mf_anc, cc_anc = None, None, None
if add_ancestors:
bp_anc, mf_anc, cc_anc = {}, {}, {}
for annot_path in annot_paths:
with open_file(annot_path) as annot_file:
for row in csv.reader(annot_file, dialect='excel-tab'):
seqid = row[0]
go_id = row[1]
if not (go_id.startswith('GO:') or go_id.startswith('EC:')):
print_stderr(
'ERROR: Unknown annotation type: {}.'.format(go_id))
continue
if go_id.startswith('EC:'):
updict_add_to_set(ec, seqid, go_id)
continue
try:
go_type = types[go_id]
except KeyError:
print_stderr(
'ERROR: Annotation not found: {}.'.format(go_id))
continue
if obsolete[go_id]:
continue
if levels[go_id] < min_level:
continue
if max_level is not None and levels[go_id] > max_level:
continue
try:
go_id = main_ids[go_id]
except KeyError:
pass
if go_type == 'biological_process':
add_annotation(bp, seqid, go_id, bp_anc, ancestors)
elif go_type == 'molecular_function':
add_annotation(mf, seqid, go_id, mf_anc, ancestors)
elif go_type == 'cellular_component':
add_annotation(cc, seqid, go_id, cc_anc, ancestors)
return bp, mf, cc, ec, bp_anc, mf_anc, cc_anc
def check_arguments(args):
errors = []
errors += check_file_arg(args.input_path, stdio_allowed=True)
errors += check_file_arg(args.output_path,
mode='w',
stdio_allowed=True,
prefix='-o/--output_path')
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
def export_removed_list(bad_records, removed_path=REMOVED_PATH):
if not bad_records:
print_stderr('No bad record found.', time=False)
else:
nb_bad = len(bad_records)
bad = '\n'.join(sorted(bad_records))
if nb_bad == 1:
msg = '1 bad record found: {}'.format(bad)
else:
msg = '{:d} bad records found:\n{}'.format(nb_bad, bad)
print_stderr(msg, time=False)
if removed_path is not None:
with open(removed_path, 'w') as removed_file:
removed_file.write(bad)
def check_arguments(args):
errors = []
errors += check_sequtils_inputs(args, stdin_allowed=True)
errors += check_file_arg(args.seq_counts_path,
mode='w',
none_allowed=True,
prefix='-S/--seq_counts_path')
errors += check_file_arg(args.base_counts_path,
mode='w',
none_allowed=True,
prefix='-C/--base_counts_path')
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
def check_arguments(args):
errors = []
errors += check_inputs(args)
errors += check_outputs(args, stdout_allowed=True)
errors += check_cpu_count(args)
errors += check_file_arg(args.removed_path,
mode='w',
none_allowed=True,
prefix='-r/--removed')
if args.no_output:
args.output_paths = None
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
def check_arguments(args):
errors = []
for annot_path in args.annot_paths:
errors += check_file_arg(annot_path, stdio_allowed=True)
errors += check_file_arg(args.tree_path, prefix='-t/--tree_path')
errors += check_file_arg(args.seqids_path,
none_allowed=True,
prefix='-s/--seqids_path')
errors += check_file_arg(args.output_path,
mode='w',
stdio_allowed=True,
prefix='-o/--output_path')
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
def check_arguments(args):
errors = []
errors += check_inputs(args, zip_ext=BGZIP_EXT)
if args.output_path is None:
if len(args.input_paths) > 1:
args.output_path = 'index.idx'
else:
input_basename = os.path.basename(args.input_paths[0])
args.output_path = input_basename + '.idx'
errors += check_file_arg(args.output_path,
mode='w',
prefix='-o/--output_path')
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
def get_search_settings(args):
seqids = get_values(args.seqids, args.seqids_path)
descriptions = get_values(args.descriptions, args.descriptions_path)
names = get_values(args.names, args.names_path)
mol_types = get_values(args.mol_types, args.mol_types_path)
taxids = get_values(args.taxids, args.taxids_path)
if not (seqids or descriptions or names or mol_types or taxids):
print_stderr('ERROR: No search criterion specified.')
sys.exit(0)
simple_case = check_if_simple_case(
seqids, descriptions, names, mol_types, taxids,
args.inverse, args.comparison)
comparison_func = set_comparison_func(args.comparison)
return (seqids, descriptions, names, mol_types, taxids,
comparison_func, simple_case)
def extract_records_by_seqid(seqids, input_paths=INPUT_PATHS,
input_format=SEQFILE_FORMAT,
index_path=INDEX_PATH, output_path=OUTPUT_PATH,
output_format=SEQFILE_FORMAT):
records_list = make_indexed_records_list(
input_paths, index_path, input_format)
with open_file(output_path, 'w') as output_file:
not_found = []
for seqid in seqids:
try:
record = get_indexed_record(seqid, records_list)
except KeyError:
not_found.append(seqid)
else:
write_records(record, output_file, output_format)
if not_found:
print_stderr('{:d} identifier(s) not found: {}'.format(
len(not_found), ', '.join(not_found)))
def check_arguments(args):
errors = []
errors += check_outputs(args, stdout_allowed=True)
errors += check_cpu_count(args)
errors += check_num_arg(
args.min_length, number_type=int, mini=1,
prefix='-l/--min_length')
if args.max_length is None:
args.max_length = args.min_length
else:
errors += check_num_arg(
args.max_length, number_type=int, mini=args.min_length,
prefix='-L/--max_length')
errors += check_num_arg(
args.nb_sequences, number_type=int, mini=1,
prefix='-n/--nb_sequences')
if args.seq_type is None:
args.seq_type = SEQ_TYPE
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
def check_arguments(args):
errors = []
errors += check_sequtils_inputs(args, stdin_allowed=True)
errors += check_sequtils_outputs(args, stdout_allowed=True)
errors += check_num_arg(args.min_length,
number_type=int,
mini=1,
prefix='-l/--min_length')
errors += check_num_arg(args.max_length,
number_type=int,
mini=args.min_length,
none_allowed=True,
prefix='-L/--max_length')
errors += check_num_arg(args.min_count,
number_type=int,
mini=0,
prefix='-n/--min_count')
errors += check_num_arg(args.max_count,
number_type=int,
mini=args.min_count,
none_allowed=True,
prefix='-N/--max_count')
errors += check_num_arg(args.min_prop,
number_type=float,
mini=0.0,
maxi=100.0,
prefix='-p/--min_prop')
errors += check_num_arg(args.max_prop,
number_type=float,
mini=args.min_prop,
maxi=100.0,
prefix='-P/--max_prop')
if args.counted_char is not None and len(args.counted_char) != 1:
errors.append('-c/--counted_char: Must be one character.')
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
def check_arguments(args):
errors = []
for annot_path in args.annot_paths:
errors += check_file_arg(annot_path, stdio_allowed=True)
for sample_path in args.sample_paths:
errors += check_file_arg(sample_path, prefix='-s/--sample_path')
errors += check_file_arg(args.tree_path, prefix='-t/--tree_path')
errors += check_file_arg(args.ref_path, prefix='-r/--ref_path')
errors += check_file_arg(args.info_path,
none_allowed=True,
prefix='-i/--info_path')
errors += check_dir_arg(args.output_dir_path,
mode='w',
create=True,
prefix='-o/--output_dir_path')
errors += check_bin_arg(args.lualatex_path, prefix='--lualatex_path')
errors += check_num_arg(args.min_level,
number_type=int,
mini=1,
prefix='-l/--min_level')
errors += check_num_arg(args.max_level,
number_type=int,
mini=1,
none_allowed=True,
prefix='-L/--max_level')
errors += check_num_arg(args.min_fold_change,
number_type=float,
mini=1.0,
prefix='-f/--min_fold_change')
errors += check_num_arg(args.max_pvalue,
number_type=float,
mini=0.0,
maxi=1.0,
prefix='-p/--max_pvalue')
errors += check_num_arg(args.min_seq_count,
number_type=int,
mini=0,
prefix='-n/--min_seq_count')
errors += check_num_arg(args.min_seq_prop,
number_type=float,
mini=0.0,
maxi=100.0,
prefix='-m/--min_seq_prop')
if args.max_level is not None and args.min_level > args.max_level:
error = '-l/-L: Max level must be greater than min level.'
errors.append(error)
if args.sample_names is not None:
if len(args.sample_names) != len(args.sample_paths):
error = ('-s/-S: The number of sample names does not match '
'the number of sample paths.')
errors.append(error)
if errors:
for error in errors:
print_stderr(error, prefix='ERROR')
sys.exit(1)
if args.output_types is None:
args.output_types = OUTPUT_TYPES
if not (args.process or args.function or args.component):
args.process, args.function, args.component = True, True, True
if not (args.up_reg or args.down_reg or args.not_reg):
args.up_reg, args.down_reg, args.not_reg = True, True, False