def test_get_matching_files(self):
""" Properly returns matching reads/barcodes/mapping files """
all_fastq = [
'sample1_r1_000.fastq', 'sample2_r1_000.fastq',
'sample1_bc_000.fastq', 'sample2_bc_000.fastq'
]
all_mapping = ['sample1_mapping_000.txt', 'sample2_mapping_000.txt']
read_indicator = 'r1'
barcode_indicator = 'bc'
mapping_indicator = 'mapping'
actual_matching_files = get_matching_files(all_fastq, all_mapping,
read_indicator,
barcode_indicator,
mapping_indicator)
actual_reads = set(actual_matching_files.keys())
actual_bcs_mapping = set(actual_matching_files.values())
expected_matching_reads = set(
['sample1_r1_000.fastq', 'sample2_r1_000.fastq'])
expected_matching_bcs_reads = set([
('sample1_bc_000.fastq', 'sample1_mapping_000.txt'),
('sample2_bc_000.fastq', 'sample2_mapping_000.txt')
])
self.assertEqual(actual_reads, expected_matching_reads)
self.assertEqual(actual_bcs_mapping, expected_matching_bcs_reads)
def test_get_matching_files(self):
""" Properly returns matching reads/barcodes/mapping files """
all_fastq = ['sample1_r1_000.fastq', 'sample2_r1_000.fastq',
'sample1_bc_000.fastq', 'sample2_bc_000.fastq']
all_mapping = ['sample1_mapping_000.txt', 'sample2_mapping_000.txt']
read_indicator = 'r1'
barcode_indicator = 'bc'
mapping_indicator = 'mapping'
actual_matching_files = get_matching_files(all_fastq, all_mapping,
read_indicator, barcode_indicator, mapping_indicator)
actual_reads = set(actual_matching_files.keys())
actual_bcs_mapping = set(actual_matching_files.values())
expected_matching_reads = set(['sample1_r1_000.fastq',
'sample2_r1_000.fastq'])
expected_matching_bcs_reads = set([('sample1_bc_000.fastq',
'sample1_mapping_000.txt'),
('sample2_bc_000.fastq', 'sample2_mapping_000.txt')])
self.assertEqual(actual_reads, expected_matching_reads)
self.assertEqual(actual_bcs_mapping, expected_matching_bcs_reads)
def main():
option_parser, opts, args =\
parse_command_line_parameters(suppress_verbose=True, **script_info)
input_dir = opts.input_dir
demultiplexing_method = opts.demultiplexing_method
parameter_fp = opts.parameter_fp
read_indicator = opts.read_indicator
barcode_indicator = opts.barcode_indicator
mapping_indicator = opts.mapping_indicator
mapping_extensions = opts.mapping_extensions.split(',')
sampleid_indicator = opts.sampleid_indicator
leading_text = opts.leading_text
trailing_text = opts.trailing_text
include_input_dir_path = opts.include_input_dir_path
output_dir = abspath(opts.output_dir)
remove_filepath_in_name = opts.remove_filepath_in_name
print_only = opts.print_only
if remove_filepath_in_name and not include_input_dir_path:
option_parser.error("If --remove_filepath_in_name enabled, "
"--include_input_dir_path must be enabled.")
if opts.parameter_fp:
with open(opts.parameter_fp, 'U') as parameter_f:
params_dict = parse_qiime_parameters(parameter_f)
params_str = get_params_str(params_dict['split_libraries_fastq'])
else:
params_dict = {}
params_str = ""
create_dir(output_dir)
all_fastq = []
all_mapping = []
extensions = ['.fastq.gz', '.fastq', '.fq.gz', '.fq']
for root, dir, fps in walk(input_dir):
for fp in fps:
for extension in extensions:
if fp.endswith(extension):
all_fastq += [abspath(join(root, fp))]
if demultiplexing_method == 'mapping_barcode_files':
for root, dir, fps in walk(input_dir):
for fp in fps:
for mapping_extension in mapping_extensions:
if fp.endswith(mapping_extension):
all_mapping += [abspath(join(root, fp))]
all_files = get_matching_files(all_fastq, all_mapping, read_indicator,
barcode_indicator, mapping_indicator)
else:
all_files = all_fastq
commands = create_commands_slf(all_files, demultiplexing_method,
output_dir, params_str, leading_text,
trailing_text, include_input_dir_path,
remove_filepath_in_name, sampleid_indicator)
qiime_config = load_qiime_config()
if print_only:
command_handler = print_commands
else:
command_handler = call_commands_serially
logger = WorkflowLogger(generate_log_fp(output_dir),
params=params_dict,
qiime_config=qiime_config)
# Call the command handler on the list of commands
command_handler(commands,
status_update_callback=no_status_updates,
logger=logger,
close_logger_on_success=True)
def main():
option_parser, opts, args =\
parse_command_line_parameters(suppress_verbose=True, **script_info)
input_dir = opts.input_dir
demultiplexing_method = opts.demultiplexing_method
parameter_fp = opts.parameter_fp
read_indicator = opts.read_indicator
barcode_indicator = opts.barcode_indicator
mapping_indicator = opts.mapping_indicator
mapping_extensions = opts.mapping_extensions.split(',')
sampleid_indicator = opts.sampleid_indicator
leading_text = opts.leading_text
trailing_text = opts.trailing_text
include_input_dir_path = opts.include_input_dir_path
output_dir = abspath(opts.output_dir)
remove_filepath_in_name = opts.remove_filepath_in_name
print_only = opts.print_only
if remove_filepath_in_name and not include_input_dir_path:
option_parser.error("If --remove_filepath_in_name enabled, "
"--include_input_dir_path must be enabled.")
if opts.parameter_fp:
with open(opts.parameter_fp, 'U') as parameter_f:
params_dict = parse_qiime_parameters(parameter_f)
params_str = get_params_str(params_dict['split_libraries_fastq'])
else:
params_dict = {}
params_str = ""
create_dir(output_dir)
all_fastq = []
all_mapping = []
extensions = ['.fastq.gz', '.fastq', '.fq.gz', '.fq']
for root, dir, fps in walk(input_dir):
for fp in fps:
for extension in extensions:
if fp.endswith(extension):
all_fastq += [abspath(join(root, fp))]
if demultiplexing_method == 'mapping_barcode_files':
for root, dir, fps in walk(input_dir):
for fp in fps:
for mapping_extension in mapping_extensions:
if fp.endswith(mapping_extension):
all_mapping += [abspath(join(root, fp))]
all_files = get_matching_files(all_fastq, all_mapping,
read_indicator, barcode_indicator, mapping_indicator)
else:
all_files = all_fastq
commands = create_commands_slf(all_files, demultiplexing_method, output_dir,
params_str, leading_text, trailing_text, include_input_dir_path,
remove_filepath_in_name, sampleid_indicator)
qiime_config = load_qiime_config()
if print_only:
command_handler = print_commands
else:
command_handler = call_commands_serially
logger = WorkflowLogger(generate_log_fp(output_dir),
params=params_dict,
qiime_config=qiime_config)
# Call the command handler on the list of commands
command_handler(commands,
status_update_callback=no_status_updates,
logger=logger,
close_logger_on_success=True)