def construct_sample_sheet(self) -> str:
"""Construct the sample sheet"""
LOG.info("Constructing sample sheet for %s", self.flowcell_id)
# Create dummy samples for the indexes that is missing
if self.run_parameters.run_type == "wgs":
self.add_dummy_samples()
else:
LOG.info("Skip adding dummy samples since run is not WGS")
self.remove_unwanted_samples()
index.adapt_indexes(
samples=self.lims_samples,
control_software_version=self.run_parameters.
control_software_version,
reagent_kit_version=self.run_parameters.reagent_kit_version,
expected_index_length=self.run_parameters.index_length,
)
sample_sheet: str = self.convert_to_sample_sheet()
LOG.info("Validating sample sheet")
get_sample_sheet(
sample_sheet=sample_sheet,
sheet_type="S2",
bcl_converter=self.bcl_converter,
)
LOG.info("Sample sheet looks fine")
return sample_sheet
def test_validate_s2_bcl2fastq_sample_sheet_duplicate_same_lane():
# GIVEN a NovaSeq S2 sample sheet with same sample duplicated in same lane
sheet_type = "S2"
bcl_converter = "bcl2fastq"
sample_sheet_string = (
"[Data]\n"
"FCID,Lane,SampleID,SampleRef,index,index2,SampleName,Control,Recipe,Operator,Project\n"
"HWHMWDMXX,1,ACC7628A68,hg19,ATTCCACACT,TGGTCTTGTT,814206,N,R1,script,814206\n"
"HWHMWDMXX,1,ACC7628A1,hg19,AGTTAGCTGG,GATGAGAATG,814206,N,R1,script,814206\n"
"HWHMWDMXX,1,ACC7628A1,hg19,GAGAGCTAAC,CAATACTTGC,814206,N,R1,script,814206\n"
)
# WHEN validating the sampVle sheet
with pytest.raises(SampleSheetError):
# THEN assert that a sample sheet error is raised
get_sample_sheet(
sample_sheet=sample_sheet_string, sheet_type=sheet_type, bcl_converter=bcl_converter
)
def test_missing_samples():
# GIVEN a NovaSeq S2 sample sheet with a missing samples
sheet_type = "S2"
bcl_converter = "dragen"
sample_sheet_string = (
"[Data]\n"
"FCID,Lane,Sample_ID,SampleRef,index,index2,SampleName,Control,Recipe,Operator,"
"Sample_Project\n"
)
# WHEN validating the sample sheet
with pytest.raises(SampleSheetError):
# THEN assert that a sample sheet error is raised
get_sample_sheet(
sample_sheet=sample_sheet_string,
sheet_type=sheet_type,
bcl_converter=bcl_converter,
)
def test_missing_header():
# GIVEN a NovaSeq S2 sample sheet with a missing header
sheet_type = "S2"
bcl_converter = "dragen"
sample_sheet_string = (
"[Data]\n"
"HWHMWDMXX,1,ACC7628A68,hg19,ATTCCACACT,TGGTCTTGTT,814206,N,R1,script,814206\n"
"HWHMWDMXX,1,ACC7628A1,hg19,AGTTAGCTGG,GATGAGAATG,814206,N,R1,script,814206\n"
"HWHMWDMXX,2,ACC7628A1,hg19,GAGAGCTAAC,CAATACTTGC,814206,N,R1,script,814206\n"
)
# WHEN validating the sample sheet
with pytest.raises(SampleSheetError):
# THEN assert that a sample sheet error is raised
get_sample_sheet(
sample_sheet=sample_sheet_string,
sheet_type=sheet_type,
bcl_converter=bcl_converter,
)
def test_convert_to_bcl2fastq_sheet(
novaseq_bcl2fastq_sample_sheet_object: SampleSheetCreator,
project_dir: Path):
# GIVEN a sample sheet object populated with samples
assert novaseq_bcl2fastq_sample_sheet_object.lims_samples
# WHEN converting to a sample sheet
sample_sheet: str = novaseq_bcl2fastq_sample_sheet_object.construct_sample_sheet(
)
# THEN assert a correctly formatted sample sheet was created
sample_sheet_object: SampleSheet = get_sample_sheet(
sample_sheet=sample_sheet,
sheet_type="S4",
bcl_converter=novaseq_bcl2fastq_sample_sheet_object.bcl_converter,
)
assert sample_sheet_object.samples
def test_validate_s2_bcl2fastq_sample_sheet_str():
# GIVEN a NovaSeq S2 sample sheet
sheet_type = "S2"
bcl_converter = "bcl2fastq"
sample_sheet_string = (
"[Data]\n"
"FCID,Lane,SampleID,SampleRef,index,index2,SampleName,Control,Recipe,Operator,Project\n"
"HWHMWDMXX,1,ACC7628A68,hg19,ATTCCACACT,TGGTCTTGTT,814206,N,R1,script,814206\n"
"HWHMWDMXX,1,ACC7628A1,hg19,AGTTAGCTGG,GATGAGAATG,814206,N,R1,script,814206"
)
# WHEN validating the sample sheet
sheet: SampleSheet = get_sample_sheet(
sample_sheet=sample_sheet_string, sheet_type=sheet_type, bcl_converter=bcl_converter
)
# THEN assert that the sample sheet has the correct type
assert sheet.type == sheet_type
def test_validate_s2_bcl2fastq_sample_sheet_duplicate_different_lanes():
# GIVEN a NovaSeq S2 sample sheet with same sample duplicated in different lanes
sheet_type = "S2"
bcl_converter = "bcl2fastq"
sample_sheet_string = (
"[Data]\n"
"FCID,Lane,SampleID,SampleRef,index,index2,SampleName,Control,Recipe,Operator,Project\n"
"HWHMWDMXX,1,ACC7628A68,hg19,ATTCCACACT,TGGTCTTGTT,814206,N,R1,script,814206\n"
"HWHMWDMXX,1,ACC7628A1,hg19,AGTTAGCTGG,GATGAGAATG,814206,N,R1,script,814206\n"
"HWHMWDMXX,2,ACC7628A1,hg19,GAGAGCTAAC,CAATACTTGC,814206,N,R1,script,814206\n"
)
# WHEN validating the sample sheet
sample_sheet: SampleSheet = get_sample_sheet(
sample_sheet=sample_sheet_string, sheet_type=sheet_type, bcl_converter=bcl_converter
)
# THEN assert that a sample sheet is returned with samples in it
assert sample_sheet.samples