def test_instantiate_mip_sampleinfo(sample_info_dna_raw: dict):
"""Tests sample info against a pydantic MipBaseSampleInfo"""
# GIVEN a dictionary with the basic config
# WHEN instantiating a MipBaseSampleInfo object
sample_info_object = MipBaseSampleInfo(**sample_info_dna_raw)
# THEN assert that it was successfully created
assert isinstance(sample_info_object, MipBaseSampleInfo)
def test_mip_sampleinfo_rank_model_version(sample_info_dna_raw: dict):
"""Test rank_model_version validator"""
# GIVEN a dictionary with the sample info data
# WHEN instantiating a MipBaseSampleInfo object
sample_info_object = MipBaseSampleInfo(**sample_info_dna_raw)
# THEN assert that rank_model_version was set
assert sample_info_object.rank_model_version == "v1.0"
def test_mip_sampleinfo_is_finished(sample_info_dna_raw: dict):
"""Test is_finished validator"""
# GIVEN a dictionary with the sample info data
# WHEN instantiating a MipBaseSampleInfo object
sample_info_object = MipBaseSampleInfo(**sample_info_dna_raw)
# THEN assert that is_finished was set
assert sample_info_object.is_finished
def test_mip_sampleinfo_case_id(sample_info_dna_raw: dict):
"""Test case_id validator"""
# GIVEN a dictionary with the sample info data
# WHEN instantiating a MipBaseSampleInfo object
sample_info_object = MipBaseSampleInfo(**sample_info_dna_raw)
# THEN assert that case_id was set
assert sample_info_object.case_id == "yellowhog"
def test_mip_sampleinfo_genome_build(sample_info_dna_raw: dict):
"""Test genome_build validator"""
# GIVEN a dictionary with the sample info data
# WHEN instantiating a MipBaseSampleInfo object
sample_info_object = MipBaseSampleInfo(**sample_info_dna_raw)
# THEN assert that genome build was set
assert sample_info_object.genome_build == "grch37"
def test_mip_sampleinfo_is_finished(sample_info_dna_raw: dict):
"""Test is_finished validator"""
# GIVEN a dictionary with the sample info data when analysisrunstatus is not finished
sample_info_dna_raw["analysisrunstatus"] = "not_finished"
# WHEN instantiating a MipBaseSampleInfo object
sample_info_object = MipBaseSampleInfo(**sample_info_dna_raw)
# THEN assert that is_finished was set
assert not sample_info_object.is_finished
def test_mip_sampleinfo_case_id_with_family_id(sample_info_dna_raw: dict):
"""Test case_id validator"""
# GIVEN a MIP sample info file missing a case_id but with family_id
sample_info_dna_raw.pop("case_id")
# WHEN instantiating a MipBaseSampleInfo object
sample_info_object = MipBaseSampleInfo(**sample_info_dna_raw)
# THEN assert that case_id was set
assert sample_info_object.case_id == "a_family_id"
def test_mip_sampleinfo(case_qc_sample_info_path: Path):
"""Test to parse the content of a real qc_sample_info file"""
# GIVEN the path to a file with sample_info metadata content
with open(case_qc_sample_info_path, "r") as sample_info_handle:
raw_sample_info = yaml.full_load(sample_info_handle)
# WHEN instantiating a MipBaseSampleInfo object
sample_info_object = MipBaseSampleInfo(**raw_sample_info)
# THEN assert that it was successfully created
assert isinstance(sample_info_object, MipBaseSampleInfo)
def parse_mip_analysis(mip_config_raw: dict, qc_metrics_raw: dict,
sample_info_raw: dict) -> MipAnalysis:
"""Parse the output analysis files from MIP
Args:
mip_config_raw (dict): raw YAML input from MIP analysis config file
qc_metrics_raw (dict): raw YAML input from MIP analysis qc metric file
sample_info_raw (dict): raw YAML input from MIP analysis qc sample info file
Returns:
MipAnalysis: parsed MIP analysis data
"""
mip_config: MipBaseConfig = MipBaseConfig(**mip_config_raw)
qc_metrics = MIPMetricsDeliverables(**qc_metrics_raw)
sample_info: MipBaseSampleInfo = MipBaseSampleInfo(**sample_info_raw)
return MipAnalysis(
case=mip_config.case_id,
genome_build=sample_info.genome_build,
sample_id_metrics=qc_metrics.sample_id_metrics,
mip_version=sample_info.mip_version,
rank_model_version=sample_info.rank_model_version,
sample_ids=mip_config.sample_ids,
sv_rank_model_version=sample_info.sv_rank_model_version,
)
def get_pipeline_version(self, case_id: str) -> str:
"""Get MIP version from sample info file"""
LOG.debug("Fetch pipeline version")
sample_info_raw = yaml.safe_load(self.get_sample_info_path(case_id).open())
sample_info: MipBaseSampleInfo = MipBaseSampleInfo(**sample_info_raw)
return sample_info.mip_version
def fixture_sample_info_dna(sample_info_dna_raw: dict) -> MipBaseSampleInfo:
return MipBaseSampleInfo(**sample_info_dna_raw)