def main():
option_parser, opts, args = \
parse_command_line_parameters(suppress_verbose=True, **script_info)
input_dir = opts.input_dir
parameter_fp = opts.parameter_fp
read1_indicator = opts.read1_indicator
read2_indicator = opts.read2_indicator
match_barcodes = opts.match_barcodes
barcode_indicator = opts.barcode_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 is enabled, "
"--include_input_dir_path must also 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['join_paired_ends'])
else:
params_dict = {}
params_str = ""
create_dir(output_dir)
all_files = []
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_files += [abspath(join(root, fp))]
pairs, bc_pairs = get_pairs(all_files, read1_indicator, read2_indicator,
match_barcodes, barcode_indicator)
commands = create_commands_jpe(pairs, output_dir,
params_str, leading_text, trailing_text, include_input_dir_path,
remove_filepath_in_name, match_barcodes, bc_pairs)
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
paired_data = opts.paired_data
parameter_fp = opts.parameter_fp
read1_indicator = opts.read1_indicator
read2_indicator = opts.read2_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 is enabled, "
"--include_input_dir_path must also 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['extract_barcodes'])
else:
params_dict = {}
params_str = ""
create_dir(output_dir)
all_files = []
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_files += [abspath(join(root, fp))]
if paired_data:
all_files, bc_pairs = get_pairs(all_files, read1_indicator,
read2_indicator)
commands = create_commands_eb(all_files, paired_data, output_dir,
params_str, leading_text, trailing_text, include_input_dir_path,
remove_filepath_in_name)
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 test_get_pairs(self):
""" Properly returns pairs of matching fastq forward/reverse reads """
read1_indicator = "_R1_"
read2_indicator = "_R2_"
input_forward_reads = ['/home/reads/read1_R1_000.fastq',
'/home/reads/read1_R2_000.fastq', '/home/not_a_matched_read/_R1_']
actual_paired_reads, actual_bc_reads =\
get_pairs(input_forward_reads, read1_indicator, read2_indicator)
expected_paired_reads =\
{'/home/reads/read1_R1_000.fastq':'/home/reads/read1_R2_000.fastq'}
expected_bc_reads = {}
self.assertEqual(actual_paired_reads, expected_paired_reads)
self.assertEqual(actual_bc_reads, expected_bc_reads)
def test_get_pairs(self):
""" Properly returns pairs of matching fastq forward/reverse reads """
read1_indicator = "_R1_"
read2_indicator = "_R2_"
input_forward_reads = ['/home/reads/read1_R1_000.fastq',
'/home/reads/read1_R2_000.fastq', '/home/not_a_matched_read/_R1_']
actual_paired_reads, actual_bc_reads =\
get_pairs(input_forward_reads, read1_indicator, read2_indicator)
expected_paired_reads =\
{'/home/reads/read1_R1_000.fastq':'/home/reads/read1_R2_000.fastq'}
expected_bc_reads = {}
self.assertEqual(actual_paired_reads, expected_paired_reads)
self.assertEqual(actual_bc_reads, expected_bc_reads)
def test_get_pairs_bcs(self):
""" Properly returns pairs of matching reads plus barcodes """
read1_indicator = "_R1_"
read2_indicator = "_R2_"
input_forward_reads = ['/home/reads/read1_R1_000.fastq',
'/home/reads/read1_R2_000.fastq', '/home/not_a_matched_read/_R1_',
'/home/reads/read1_I1_000.fastq']
actual_paired_reads, actual_bc_reads =\
get_pairs(input_forward_reads, read1_indicator, read2_indicator,
match_barcodes = True)
expected_paired_reads =\
{'/home/reads/read1_R1_000.fastq':'/home/reads/read1_R2_000.fastq'}
expected_bc_reads = \
{'/home/reads/read1_R1_000.fastq':'/home/reads/read1_I1_000.fastq'}
self.assertEqual(actual_paired_reads, expected_paired_reads)
self.assertEqual(actual_bc_reads, expected_bc_reads)
def test_get_pairs_bcs(self):
""" Properly returns pairs of matching reads plus barcodes """
read1_indicator = "_R1_"
read2_indicator = "_R2_"
input_forward_reads = ['/home/reads/read1_R1_000.fastq',
'/home/reads/read1_R2_000.fastq', '/home/not_a_matched_read/_R1_',
'/home/reads/read1_I1_000.fastq']
actual_paired_reads, actual_bc_reads =\
get_pairs(input_forward_reads, read1_indicator, read2_indicator,
match_barcodes = True)
expected_paired_reads =\
{'/home/reads/read1_R1_000.fastq':'/home/reads/read1_R2_000.fastq'}
expected_bc_reads = \
{'/home/reads/read1_R1_000.fastq':'/home/reads/read1_I1_000.fastq'}
self.assertEqual(actual_paired_reads, expected_paired_reads)
self.assertEqual(actual_bc_reads, expected_bc_reads)