def test_validate_sample_sheet_no_data_header(self, mock_csv_reader):
"""
Given a sample sheet with no header, make sure the correct errors are included in the response
:param mock_csv_reader:
:return:
"""
field_values = ("Sample_Name,Project_ID,File_Forward,File_Reverse\n"
"my-sample-1, 72,file_1.fastq.gz,file_2.fastq.gz\n")
file_contents_str = ("{field_values}").format(
field_values=field_values)
# converts string as a pseudo file / memory file
sample_sheet_file = StringIO(file_contents_str)
# the call to get_csv_reader() inside parse_samples() will return
# items inside side_effect
mock_csv_reader.side_effect = [reader(sample_sheet_file)]
res = validate_sample_sheet(None)
# This should be an invalid sample sheet
self.assertFalse(res.is_valid())
# Only should have 2 error
self.assertEqual(len(res.error_list), 2)
# Error type should be SampleSheetError
self.assertEqual(type(res.error_list[0]), SampleSheetError)
self.assertEqual(type(res.error_list[1]), SampleSheetError)
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 test_validate_sample_sheet_no_data(self, mock_csv_reader):
"""
Given a sample sheet with no data, make sure the correct errors are included in the response
:param mock_csv_reader:
:return:
"""
file_contents_str = "[Data]\n"
# converts string as a pseudo file / memory file
sample_sheet_file = StringIO(file_contents_str)
# the call to get_csv_reader() inside parse_samples() will return
# items inside side_effect
mock_csv_reader.side_effect = [reader(sample_sheet_file)]
res = validate_sample_sheet(None)
# This should be an invalid sample sheet
self.assertFalse(res.is_valid())
# Only should have 1 error
self.assertEqual(len(res.error_list), 1)
# Error type should be SampleSheetError
self.assertEqual(type(res.error_list[0]), SampleSheetError)
def test_validate_sample_sheet_valid(self, mock_csv_reader):
"""
Given a valid sample sheet, make sure the response shows as valid
:param mock_csv_reader:
:return:
"""
field_values = ("[Data]\n"
"Sample_Name,Project_ID,File_Forward,File_Reverse\n"
"my-sample-1,72,file_1.fastq.gz,file_2.fastq.gz\n")
file_contents_str = ("{field_values}").format(
field_values=field_values)
# converts string as a pseudo file / memory file
sample_sheet_file = StringIO(file_contents_str)
# the call to get_csv_reader() inside parse_samples() will return
# items inside side_effect
mock_csv_reader.side_effect = [reader(sample_sheet_file)]
res = validate_sample_sheet(None)
# This should be a valid sample sheet
self.assertTrue(res.is_valid())
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