def get_sequencing_run(sample_sheet, run_data_directory_file_list=None):
"""
Does local validation on the integrity of the run directory / sample sheet
Throws a ValidationError with a validation result attached if it cannot make a sequencing run
:param sample_sheet:
:return: SequencingRun
"""
# get file list
validation_result = model.ValidationResult()
try:
if run_data_directory_file_list is None:
data_dir = os.path.dirname(sample_sheet)
run_data_directory_file_list = common.get_file_list(data_dir)
except exceptions.DirectoryError as error:
validation_result.add_error(error)
logging.error("Errors occurred while parsing files")
raise exceptions.ValidationError(
"Errors occurred while parsing files", validation_result)
# Try to get the sample sheet, validate that the sample sheet is valid
validation_result = validation.validate_sample_sheet(sample_sheet)
if not validation_result.is_valid():
logging.error("Errors occurred while getting sample sheet")
raise exceptions.ValidationError(
"Errors occurred while getting sample sheet",
validation_result)
# Try to build sequencing run from sample sheet & meta data, raise validation error if errors occur
try:
sample_list = sample_parser.parse_sample_list(
sample_sheet, run_data_directory_file_list)
run_metadata = sample_parser.parse_metadata(sample_list)
sequencing_run = common.build_sequencing_run_from_samples(
sample_list, run_metadata)
except exceptions.SequenceFileError as error:
validation_result.add_error(error)
logging.error(
"Errors occurred while building sequence run from sample sheet"
)
raise exceptions.ValidationError(
"Errors occurred while building sequence run from sample sheet",
validation_result)
return sequencing_run
def get_sequencing_run(self,
sample_sheet,
run_data_directory=None,
run_data_directory_file_list=None):
"""
Does local validation on the integrety of the run directory / sample sheet
Throws a ValidationError with a valadation result attached if it cannot make a sequencing run
:param sample_sheet: Sample Sheet File
:param run_data_directory: Optional: Directory (including run directory) to data files.
Can be provided for bypassing os calls when developing on cloud systems
:param run_data_directory_file_list: Optional: List of files in data directory.
Can be provided for bypassing os calls when developing on cloud systems
:return: SequencingRun
"""
# get data directory and file list
validation_result = model.ValidationResult()
try:
if run_data_directory is None:
run_data_directory = Parser.get_full_data_directory(
sample_sheet)
if run_data_directory_file_list is None:
run_data_directory_file_list = common.get_file_list(
run_data_directory)
except exceptions.DirectoryError as error:
validation_result.add_error(error)
logging.error("Errors occurred while parsing files")
raise exceptions.ValidationError(
"Errors occurred while parsing files", validation_result)
# Try to get the sample sheet, validate that the sample sheet is valid
validation_result = validation.validate_sample_sheet(sample_sheet)
if not validation_result.is_valid():
logging.error("Errors occurred while getting sample sheet")
raise exceptions.ValidationError(
"Errors occurred while getting sample sheet",
validation_result)
# Try to parse the meta data from the sample sheet, throw validation error if errors occur
validation_result = model.ValidationResult()
try:
run_metadata = sample_parser.parse_metadata(sample_sheet)
except exceptions.SampleSheetError as error:
validation_result.add_error(error)
logging.error("Errors occurred while parsing metadata")
raise exceptions.ValidationError(
"Errors occurred while parsing metadata", validation_result)
# Try to build sequencing run from sample sheet & meta data, raise validation error if errors occur
try:
sample_list = sample_parser.parse_sample_list(
sample_sheet, run_data_directory, run_data_directory_file_list)
sequencing_run = common.build_sequencing_run_from_samples(
sample_list, run_metadata, self.get_parser_type_name())
except exceptions.SequenceFileError as error:
validation_result.add_error(error)
logging.error(
"Errors occurred while building sequence run from sample sheet"
)
raise exceptions.ValidationError(
"Errors occurred while building sequence run from sample sheet",
validation_result)
return sequencing_run
def get_sequencing_run(self,
sample_sheet,
run_data_directory_file_list=None):
"""
Does local validation on the integrity of the run directory / sample sheet
Throws a ValidationError with a validation result attached if it cannot make a sequencing run
:param sample_sheet:
:param run_data_directory_file_list: Optional: List of files in the data directory to verify against the
SampleList.csv file. This is used when deploying the parsers on a cloud environment.
:return: SequencingRun
"""
# Try to get the sample sheet, validate that the sample sheet is valid
validation_result = validation.validate_sample_sheet(sample_sheet)
if not validation_result.is_valid():
logging.error("Errors occurred while getting sample sheet")
raise exceptions.ValidationError(
"Errors occurred while getting sample sheet",
validation_result)
# When running with a premade file list, verify files on sample_sheet are in file list
try:
if run_data_directory_file_list is not None:
sample_parser.verify_sample_sheet_file_names_in_file_list(
sample_sheet, run_data_directory_file_list)
except (exceptions.SequenceFileError,
exceptions.SampleSheetError) as error:
validation_result.add_error(error)
logging.error(
"Errors occurred while building sequence run from sample sheet"
)
raise exceptions.ValidationError(
"Errors occurred while building sequence run from sample sheet",
validation_result)
except Exception as error:
validation_result.add_error(error)
logging.error("System error while building sequencing run")
raise exceptions.ValidationError(
"System error while building sequencing run",
validation_result)
# Build a list of sample objects from sample sheet
try:
if run_data_directory_file_list is not None:
sample_list = sample_parser.build_sample_list_from_sample_sheet_no_verify(
sample_sheet)
else:
sample_list = sample_parser.build_sample_list_from_sample_sheet_with_abs_path(
sample_sheet)
except (exceptions.DirectoryError,
exceptions.SampleSheetError) as error:
validation_result.add_error(error)
logging.error("Errors occurred while parsing files")
raise exceptions.ValidationError(
"Errors occurred while parsing files", validation_result)
except Exception as error:
validation_result.add_error(error)
logging.error("System error while parsing files")
raise exceptions.ValidationError(
"System error while parsing files", validation_result)
# verify samples in sample_list are all of one type, either single or paired end
if not sample_parser.only_single_or_paired_in_sample_list(sample_list):
e = exceptions.SampleSheetError((
"Your sample sheet is malformed. "
"SampleSheet cannot have both paired end and single end runs. "
"Make sure all samples are either paired or single."),
sample_sheet)
validation_result.add_error(e)
logging.error(
"Error occurred while building file list: Sample sheet has both paired and single end reads"
)
raise exceptions.ValidationError(
"Errors occurred while building file list.", validation_result)
# Try to build sequencing run from sample sheet & meta data, raise validation error if errors occur
try:
run_metadata = sample_parser.parse_metadata(sample_list)
sequencing_run = common.build_sequencing_run_from_samples(
sample_list, run_metadata, self.get_parser_type_name())
except exceptions.SequenceFileError as error:
validation_result.add_error(error)
logging.error(
"Errors occurred while building sequence run from sample sheet"
)
raise exceptions.ValidationError(
"Errors occurred while building sequence run from sample sheet",
validation_result)
return sequencing_run