def test_validate_sample_sheet_missing_data_header(self, mock_csv_reader):
"""
Given a sample sheet with no data header, make sure the correct errors are included in the response
:param mock_csv_reader:
:return:
"""
h_field_values = (
"Local Run Manager Analysis Id, 4004\n"
"Experiment Name, Some_Test_Data\n"
"Date, 2015-05-14\n"
"Workflow, GenerateFastQWorkflow\n"
"Description, 12-34\n"
"Chemistry, Yes\n"
)
reads = (
"151\n"
"151\n"
)
d_headers = ("Sample_Name,"
"I7_Index_ID,index,I5_Index_ID,index2,Sample_Project,")
d_field_values = (
"15-0318-4004,15-0318,N701,TAAGGCGA,S502,CTCTCTAT,203\n"
"15-0455-4004,15-0455,N701,TAAGGCGA,S503,TATCCTCT,203\n"
"15-0462-4004,15-0462,N701,TAAGGCGA,S505,GTAAGGAG,203\n"
)
file_contents_str = (
"[Header]\n"
"{h_field_values}\n"
"[Reads]\n"
"{reads}\n"
"[Data]\n"
"{d_headers}\n"
"{d_field_values}"
).format(h_field_values=h_field_values, reads=reads, d_headers=d_headers, d_field_values=d_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 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, test that everything shows as valid
:param mock_csv_reader:
:return:
"""
h_field_values = (
"Local Run Manager Analysis Id, 4004\n"
"Experiment Name, Some_Test_Data\n"
"Date, 2015-05-14\n"
"Workflow, GenerateFastQWorkflow\n"
"Description, 12-34\n"
"Chemistry, Yes\n"
)
reads = (
"151\n"
"151\n"
)
d_headers = ("Sample_ID,Sample_Name,"
"I7_Index_ID,index,I5_Index_ID,index2,Sample_Project,")
d_field_values = (
"15-0318-4004,15-0318,N701,TAAGGCGA,S502,CTCTCTAT,203\n"
"15-0455-4004,15-0455,N701,TAAGGCGA,S503,TATCCTCT,203\n"
"15-0462-4004,15-0462,N701,TAAGGCGA,S505,GTAAGGAG,203\n"
)
file_contents_str = (
"[Header]\n"
"{h_field_values}\n"
"[Reads]\n"
"{reads}\n"
"[Data]\n"
"{d_headers}\n"
"{d_field_values}"
).format(h_field_values=h_field_values, reads=reads, d_headers=d_headers, d_field_values=d_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 test_validate_sample_sheet_no_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:
"""
headers = ("Sample_ID,Sample_Name,"
"I7_Index_ID,index,I5_Index_ID,index2,Sample_Project")
field_values = (
"15-0318-4004,15-0318,N701,TAAGGCGA,S502,CTCTCTAT,203\n"
"15-0455-4004,15-0455,N701,TAAGGCGA,S503,TATCCTCT,203\n"
"15-0462-4004,15-0462,N701,TAAGGCGA,S505,GTAAGGAG,203\n"
)
reads = (
"151\n"
"151\n"
)
file_contents_str = (
"[Reads]\n"
"{reads}\n"
"[Data]\n"
"{headers}\n"
"{field_values}"
).format(headers=headers, reads=reads, 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 1 error
self.assertEqual(len(res.error_list), 1)
# Error type should be SampleSheetError
self.assertEqual(type(res.error_list[0]), SampleSheetError)
def get_sequencing_run(sample_sheet,
run_data_directory=None,
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: 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)
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