def test_bad_records(self):
expected_messages = {
"sample_type": "Line 2: Invalid sample type: 'invalid_type'",
"experiment_id": "Line 3: Mismatch in 'experiment_id' value",
"sample_id": "Line 3: Mismatch in 'sample_id' value",
"barcode":
"Line 2: Bad 'barcode' name 'bad'; expected a name like 'barcode01'",
"internal_barcode":
"Line 2: Bad 'internal_barcode' name 'bad'; expected a name like 'internal01'",
"external_barcode":
"Line 2: Bad 'external_barcode' name 'bad'; expected a name like 'external01'",
"rapid_barcode":
"Line 2: Bad 'rapid_barcode' name 'bad'; expected a name like 'barcode01'",
"fip_barcode":
"Line 2: Bad 'fip_barcode' name 'bad'; expected a name like 'FIP01'",
"duplicate_barcode": "Line 3: Duplicate barcode: 'barcode01'",
"duplicate_alias": "Line 3: Duplicate alias: 'alias'",
}
for filename, message in expected_messages.items():
with self.assertRaisesRegex(
SampleSheetParseError,
"^" + re.escape(message) + "$",
msg="Test file: " + filename,
):
load_sample_sheet_csv(
sample_sheet_csv_path("bad_records", filename))
def test_bad_column_names(self):
expected_messages = {
"no_columns":
"No columns in sample sheet",
"invalid_columns":
"Unrecognised columns in sample sheet: 'a', 'b'",
"duplicate_columns":
"Duplicate columns in sample sheet: 'sample_id'",
"no_position_information":
"Invalid position information in sample sheet. Must have exactly one of 'flow_cell_id' and 'position_id'",
"both_position_information":
"Invalid position information in sample sheet. Must have exactly one of 'flow_cell_id' and 'position_id'",
"conflicting_barcoding_columns":
"Conflicting barcode column names: 'barcode', 'rapid_barcode', 'fip_barcode'",
"missing_barcoding_column":
"Missing barcoding column names: 'external_barcode'",
"barcoding_no_alias":
"Missing 'alias' column",
"alias_no_barcoding":
"'alias' column supplied, but no other barcode information in sample sheet",
"type_no_barcoding":
"'type' column supplied, but no other barcode information in sample sheet",
"no_user_info_columns":
"Sample sheet contains no user info. " +
"Should contain 'sample_id', 'experiment_id' or some barcode information",
}
for filename, message in expected_messages.items():
with self.assertRaisesRegex(
SampleSheetParseError,
"^" + re.escape(message) + "$",
msg="Test file: " + filename,
):
load_sample_sheet_csv(
sample_sheet_csv_path("bad_column_names", filename),
exception_on_unknown_field=True,
)
if filename != "invalid_columns":
with self.assertRaisesRegex(
SampleSheetParseError,
"^" + re.escape(message) + "$",
msg="Test file: " + filename,
):
load_sample_sheet_csv(
sample_sheet_csv_path("bad_column_names", filename),
exception_on_unknown_field=False,
)
def test_multiple_positions(self):
sample_sheet = load_sample_sheet_csv(
sample_sheet_csv_path("good", "multiple_positions"))
self.compare_sample_sheet(
sample_sheet,
[
ParsedSampleSheetEntry(
flow_cell_id=None,
position_id="A1",
sample_id="my_sample_1",
experiment_id="my_experiment",
barcode_info=[
BarcodeUserData(barcode_name="barcode01",
alias="alias01"),
],
),
ParsedSampleSheetEntry(
flow_cell_id=None,
position_id="A2",
sample_id="my_sample_2",
experiment_id="my_experiment",
barcode_info=[
BarcodeUserData(barcode_name="barcode01",
alias="alias01"),
],
),
],
)
def test_lampore_barcoding(self):
sample_sheet = load_sample_sheet_csv(
sample_sheet_csv_path("good", "lampore_barcoding"))
self.compare_sample_sheet(
sample_sheet,
[
ParsedSampleSheetEntry(
flow_cell_id="FC001",
position_id=None,
sample_id=None,
experiment_id=None,
barcode_info=[
BarcodeUserData(
barcode_name="barcode01",
lamp_barcode_id="FIP01",
alias="alias01",
),
BarcodeUserData(
barcode_name="barcode02",
lamp_barcode_id="FIP01",
alias="alias02",
),
],
),
],
)
def test_no_error_on_unknown_fields(self):
sample_sheet = load_sample_sheet_csv(
sample_sheet_csv_path("bad_column_names", "invalid_columns"),
exception_on_unknown_field=False,
)
# Should have an empty sample sheet, since no entries
self.assertFalse(sample_sheet)
def test_no_barcoding(self):
sample_sheet = load_sample_sheet_csv(
sample_sheet_csv_path("good", "no_barcoding"))
self.compare_sample_sheet(
sample_sheet,
[
ParsedSampleSheetEntry(
flow_cell_id="FC001",
position_id=None,
sample_id="my_sample",
experiment_id="my_experiment",
barcode_info=None,
),
],
)
def test_single_barcoding(self):
sample_sheet = load_sample_sheet_csv(
sample_sheet_csv_path("good", "single_barcoding"))
SampleType = BarcodeUserData.SampleType
self.compare_sample_sheet(
sample_sheet,
[
ParsedSampleSheetEntry(
flow_cell_id="FC001",
position_id=None,
sample_id=None,
experiment_id=None,
barcode_info=[
BarcodeUserData(
barcode_name="barcode01",
alias="alias01",
type=SampleType.test_sample,
),
BarcodeUserData(
barcode_name="barcode02",
alias="alias02",
type=SampleType.positive_control,
),
BarcodeUserData(
barcode_name="barcode03",
alias="alias03",
type=SampleType.negative_control,
),
BarcodeUserData(
barcode_name="barcode04",
alias="alias04",
type=SampleType.no_template_control,
),
],
),
],
)
def add_sample_sheet_entries(experiment_specs: ExperimentSpecs, args):
if args.sample_sheet:
assert not args.position
assert not args.flow_cell_id
try:
sample_sheet = load_sample_sheet_csv(args.sample_sheet)
except SampleSheetParseError as e:
print("Error loading sample sheet CSV: {}".format(e))
sys.exit(1)
except FileNotFoundError as e:
print("Sample sheet file not found: {}".format(args.sample_sheet))
sys.exit(1)
for entry in sample_sheet:
# Fix up the `sample_id` and `experiment_id` with values supplied on the command line
# (provided they don't clash with sample sheet values)
if args.sample_id:
if entry.sample_id and args.sample_id != entry.sample_id:
print(
"`--sample-id` specified on command line does not match `sample_id` specified in sample sheet"
)
sys.exit(1)
else:
sample_id = args.sample_id
else:
sample_id = entry.sample_id
if args.experiment_group:
if entry.experiment_id and args.experiment_group != entry.experiment_id:
print(
"`--experiment-group` specified on command line does not match `experiment_id` specified in sample sheet"
)
sys.exit(1)
else:
experiment_id = args.experiment_group
else:
experiment_id = entry.experiment_id
# Check that we have exactly one of `position_id` and `flow_cell_id`
assert not (entry.position_id and entry.flow_cell_id)
# Add the entry to the specs
experiment_specs.append(
ExperimentSpec(entry=ParsedSampleSheetEntry(
flow_cell_id=entry.flow_cell_id,
position_id=entry.position_id,
sample_id=sample_id,
experiment_id=experiment_id,
barcode_info=entry.barcode_info,
)))
elif args.position:
assert not args.sample_sheet
assert not args.flow_cell_id
experiment_specs.append(
ExperimentSpec(entry=ParsedSampleSheetEntry(
flow_cell_id=None,
position_id=args.position,
sample_id=args.sample_id,
experiment_id=args.experiment_group,
barcode_info=None,
)))
elif args.flow_cell_id:
assert not args.sample_sheet
assert not args.position
experiment_specs.append(
ExperimentSpec(entry=ParsedSampleSheetEntry(
flow_cell_id=args.flow_cell_id,
position_id=None,
sample_id=args.sample_id,
experiment_id=args.experiment_group,
barcode_info=None,
)))
else:
print(
"No sample position specified; specify either `--position`, `--flow-cell-id` or `--sample-sheet`"
)
sys.exit(1)