def migrate_interpretation_request_cancer_to_interpreted_genome_v6(
json_dict, assembly, interpretation_service, reference_database_versions, software_versions,
report_url, comments):
"""
:type json_dict: dict
:type assembly: Assembly
:type interpretation_service: str
:type reference_database_versions: dict
:type software_versions: dict
:type report_url: str
:type comments: list
:rtype: CancerInterpretationRequest_6_0_0
"""
if CancerInterpretationRequest_5_0_0.validate(CancerInterpretationRequest_5_0_0.fromJsonDict(json_dict)):
raise MigrationError(
"Cannot transform a cancer interpretation request in version 5.0.0 into an interpreted genome")
if CancerInterpretationRequest_6_0_0.validate(CancerInterpretationRequest_6_0_0.fromJsonDict(json_dict)):
raise MigrationError(
"Cannot transform a cancer interpretation request in version 6.0.0 into an interpreted genome")
types = [
InterpretedGenome_6_0_0,
CancerInterpretedGenome_5_0_0,
CancerInterpretationRequest_4_0_0
]
migrations = [
lambda x: x,
MigrateReports500To600().migrate_cancer_interpreted_genome,
lambda x: MigrateReports400To500().migrate_cancer_interpretation_request_to_cancer_interpreted_genome(
old_instance=x, assembly=assembly, interpretation_service=interpretation_service,
reference_database_versions=reference_database_versions, software_versions=software_versions,
report_url=report_url, comments=comments)
]
return MigrationHelpers.migrate(json_dict, types, migrations)
def migrate_cancer_interpretation_request_to_cancer_interpreted_genome(
self,
old_instance,
assembly,
interpretation_service,
reference_database_versions,
software_versions,
report_url=None,
comments=None):
"""
NOTE: we migrate from a model where only one sample and one participant is supported, thus we do not need
a list of samples or participants
:type old_instance: reports_4_0_0.CancerInterpretationRequest
:type assembly: reports_5_0_0.Assembly
:type interpretation_service: str
:type reference_database_versions: dict
:type software_versions: dict
:type report_url: str
:type comments: list
:rtype: reports_5_0_0.CancerInterpretedGenome
"""
new_instance = self.convert_class(
self.new_model.CancerInterpretedGenome,
old_instance) # :type: reports_5_0_0.CancerInterpretedGenome
new_instance.interpretationRequestId = old_instance.reportRequestId
new_instance.interpretationRequestVersion = old_instance.reportVersion
new_instance.interpretationService = interpretation_service
new_instance.referenceDatabasesVersions = reference_database_versions
if not isinstance(software_versions, dict):
software_versions = {}
software_versions['tiering'] = old_instance.tieringVersion
new_instance.softwareVersions = software_versions
new_instance.reportUrl = report_url
new_instance.comments = comments
participant_id = old_instance.cancerParticipant.individualId
tumor_samples = old_instance.cancerParticipant.tumourSamples
if not tumor_samples:
raise MigrationError(
"There is no tumour sample to perform the migration")
elif len(tumor_samples) > 1:
raise MigrationError(
"There are several tumour samples, cannot decide which to use '{}'"
.format(str(tumor_samples)))
sample_id = tumor_samples[0].sampleId
new_instance.variants = self.convert_collection(
old_instance.tieredVariants,
self._migrate_reported_variant_cancer,
assembly=assembly,
participant_id=participant_id,
sample_id=sample_id)
return self.validate_object(
object_to_validate=new_instance,
object_type=self.new_model.CancerInterpretedGenome)
def migrate_cancer_exit_questionnaire(self, old_instance, assembly):
"""
:type old_instance: reports_5_0_0.CancerExitQuestionnaire
:type assembly: reports_5_0_0.Assembly
:rtype: reports_6_0_0.CancerExitQuestionnaire
"""
if assembly is None:
raise MigrationError(
"Parameter <assembly> is required to migrate cancer exit questionnaire to version 6"
)
new_c_eq = self.convert_class(
target_klass=self.new_model.CancerExitQuestionnaire,
instance=old_instance)
new_c_eq.somaticVariantLevelQuestions = self.convert_collection(
old_instance.somaticVariantLevelQuestions,
self._migrate_somatic_variant_level_question,
assembly=assembly)
new_c_eq.germlineVariantLevelQuestions = self.convert_collection(
old_instance.germlineVariantLevelQuestions,
self._migrate_germline_variant_level_question,
assembly=assembly)
new_c_eq.otherActionableVariants = self.convert_collection(
old_instance.otherActionableVariants,
self._migrate_other_actionable_variant,
assembly=assembly)
return self.validate_object(
object_to_validate=new_c_eq,
object_type=self.new_model.CancerExitQuestionnaire)
def migrate_interpreted_genome_rd(self, old_instance, assembly,
interpretation_request_version):
"""
:type old_instance: reports_4_0_0.InterpretedGenomeRD
:type assembly: reports_5_0_0.Assembly
:type interpretation_request_version: int
:rtype: reports_5_0_0.InterpretedGenomeRD
"""
if assembly is None or interpretation_request_version is None:
raise MigrationError(
"Parameters <assembly> and <interpretation_request_version> are required for models earlier than 5.0.0"
)
new_instance = self.convert_class(
self.new_model.InterpretedGenomeRD,
old_instance) # type:self.new_model.InterpretedGenomeRD
new_instance.interpretationRequestVersion = interpretation_request_version
new_instance.interpretationService = old_instance.companyName
new_instance.variants = self.convert_collection(
old_instance.reportedVariants,
self._migrate_reported_variant,
assembly=assembly)
return self.validate_object(
object_to_validate=new_instance,
object_type=self.new_model.InterpretedGenomeRD)
def migrate_action(self, old_instance):
"""
NOTE: fields that cannot be filled are "actionType"
:type old_instance: reports_4_0_0.Actions
:rtype reports_5_0_0.Action
:return:
"""
new_instance = self.convert_class(self.new_model.Action, old_instance)
new_instance.evidenceType = old_instance.actionType
new_instance.actionType = None
# rename evidence to references
new_instance.references = old_instance.evidence
# maps the action status
if old_instance.status is not None:
status = old_instance.status.lower().replace('-', '_')
if status == reports_5_0_0.ActionStatus.clinical:
new_instance.status = reports_5_0_0.ActionStatus.clinical
elif status == reports_5_0_0.ActionStatus.pre_clinical:
new_instance.status = reports_5_0_0.ActionStatus.pre_clinical
elif status is not None and status != "":
raise MigrationError("Action status does not match any known value '{}'".format(status))
return self.validate_object(
object_to_validate=new_instance, object_type=self.new_model.Action
)
def _migrate_reported_variant_cancer(self, old_instance, assembly,
participant_id, sample_ids):
ne_instance = old_instance.reportedVariantCancer
new_instance = self.convert_class(
self.new_model.ReportedVariantCancer,
ne_instance) # :type: reports_5_0_0.ReportedVariant
new_instance.variantCoordinates = self.convert_class(
reports_5_0_0.VariantCoordinates, ne_instance)
new_instance.variantCoordinates.assembly = self._migrate_assembly(
assembly)
if old_instance.reportedVariantCancer.cDnaChange:
new_instance.cdnaChanges = [
old_instance.reportedVariantCancer.cDnaChange
]
if ne_instance.proteinChange:
new_instance.proteinChanges = [ne_instance.proteinChange]
# NOTE: missing fields: genomicChanges
sample_id = sample_ids.get(old_instance.alleleOrigins[0], None)
if not sample_id:
raise MigrationError('Couldn\'t retrieve Sample ID for {}'.format(
old_instance.alleleOrigins[0]))
# builds up the VariantCall object
# NOTE: fields that cannot be filled "phaseSet"
new_instance.variantCalls = [
reports_5_0_0.VariantCall(
depthReference=ne_instance.depthReference,
depthAlternate=ne_instance.depthAlternate,
vaf=ne_instance.vaf,
zygosity=reports_5_0_0.Zygosity.na,
alleleOrigins=old_instance.alleleOrigins,
participantId=participant_id,
sampleId=sample_id)
]
if ne_instance.commonAf is not None:
new_instance.alleleFrequencies = [
reports_5_0_0.AlleleFrequency(
study='genomics_england',
population='ALL',
alternateFrequency=self.convert_string_to_float(
ne_instance.commonAf) / 100)
]
# NOTE: some fields cannot be filled: "fdp50", "recurrentlyReported", "others"
new_instance.variantAttributes = reports_5_0_0.VariantAttributes(
ihp=ne_instance.ihp)
new_instance.alleleOrigins = old_instance.alleleOrigins
new_instance.reportEvents = self.convert_collection(
list(zip(ne_instance.reportEvents, new_instance.reportEvents)),
self._migrate_report_event_cancer)
return new_instance
def _migrate_assembly(self, assembly):
new_assembly = None
if assembly is not None:
if assembly.lower().startswith(reports_5_0_0.Assembly.GRCh37.lower()) \
or assembly.lower().startswith('hg19'):
new_assembly = reports_5_0_0.Assembly.GRCh37
elif assembly.lower().startswith(
reports_5_0_0.Assembly.GRCh38.lower()):
new_assembly = reports_5_0_0.Assembly.GRCh38
else:
raise MigrationError(
"Assembly does not match any known value '{}'".format(
assembly))
return new_assembly
def _extract_variant_details(variant_details):
"""
The format of variant_details is "chr:pos:ref:alt"
"""
details = list(
map(lambda x: x.strip(),
re.compile(":|>").split(variant_details)))
if len(details) != 4:
raise MigrationError(
"Variant details: {variant_details} should have fields chr, pos, ref and alt"
.format(variant_details=variant_details))
try:
details[1] = int(details[1])
except ValueError:
raise MigrationError(
"Position {position} is not an integer !".format(
position=details[1]))
return {
"chromosome": details[0],
"position": details[1],
"reference": details[2],
"alternate": details[3],
}
def migrate_interpretation_request_rd(self, old_instance, old_ig):
"""
Migrates a reports_5_0_0.InterpretationRequestRD into a reports_4_0_0.InterpretationRequestRD
:type old_instance: reports_5_0_0.InterpretationRequestRD
:type old_ig: reports_5_0_0.InterpretedGenomeRD
:rtype: reports_4_0_0.InterpretationRequestRD
"""
new_instance = self.convert_class(
self.new_model.InterpretationRequestRD, old_instance)
new_instance.versionControl = self.new_model.ReportVersionControl()
new_instance.genomeAssemblyVersion = old_instance.genomeAssembly
# ensure null lists of files are not passing through
if new_instance.bams is None:
new_instance.bams = []
if new_instance.vcfs is None:
new_instance.vcfs = []
# grabs the list of variants from the interpreted genome
new_instance.tieredVariants = self.convert_collection(
old_ig.variants, self._migrate_reported_variant)
new_instance.tieringVersion = old_ig.softwareVersions.get(
"tiering", "")
new_instance.analysisVersion = ''
new_instance.analysisReturnUri = ''
if old_instance.additionalInfo:
new_instance.analysisVersion = old_instance.additionalInfo.get(
'analysisVersion') or ''
new_instance.analysisReturnUri = old_instance.additionalInfo.get(
'analysisReturnUri', '')
new_instance.tieringVersion = old_instance.additionalInfo.get(
'tieringVersion', '')
new_instance.complexGeneticPhenomena = old_instance.additionalInfo.get(
'complexGeneticPhenomena')
new_instance.cellbaseVersion = old_instance.additionalInfo.get(
'cellbaseVersion', '')
new_instance.interpretGenome = bool(
distutils.util.strtobool(
old_instance.additionalInfo.get('interpretGenome',
'false')))
if not old_instance.pedigree:
raise MigrationError(
"Cannot reverse migrate an Interpretation Request for RD with null pedigree"
)
new_instance.pedigree = MigrateParticipant110To100().migrate_pedigree(
old_instance.pedigree)
return self.validate_object(
object_to_validate=new_instance,
object_type=self.new_model.InterpretationRequestRD)
def _migrate_action(self, actions):
old_instance = actions[0]
new_instance = actions[1]
new_instance.evidenceType = old_instance.actionType
new_instance.actionType = None
new_instance.references = old_instance.evidence
if old_instance.status is not None:
status = old_instance.status.lower().replace('-', '_')
if status == reports_5_0_0.ActionStatus.clinical:
new_instance.status = reports_5_0_0.ActionStatus.clinical
elif status == reports_5_0_0.ActionStatus.pre_clinical:
new_instance.status = reports_5_0_0.ActionStatus.pre_clinical
elif status is not None and status != "":
raise MigrationError(
"Action status does not match any known value '{}'".format(
status))
return new_instance
def migrate_tumour_sample(self, old_sample, ldp_code):
ts = self.new_model.TumourSample
tt = self.new_model.TumourType
new_sample = self.convert_class(ts, old_sample) # :type: reports_5_0_0.TumourSample
tumour_type_enum = [
tt.PRIMARY, tt.METASTATIC_RECURRENCE, tt.RECURRENCE_OF_PRIMARY_TUMOUR, tt.METASTASES
]
new_sample.tumourType = old_sample.tumourType if old_sample.tumourType in tumour_type_enum else None
new_sample.tumourId = self.convert_int_to_str(value=old_sample.tumourId)
if ldp_code is None:
raise MigrationError("Cannot migrate '{}' to '{}' having an empty value of 'ldp_code'".format(
type(old_sample), type(new_sample))
)
new_sample.LDPCode = ldp_code
return self.validate_object(object_to_validate=new_sample, object_type=ts)
def migrate_rd_exit_questionnaire(self, old_instance, assembly):
"""
:type old_instance: reports_5_0_0.RareDiseaseExitQuestionnaire
:type assembly: reports_5_0_0.Assembly
:rtype: reports_6_0_0.RareDiseaseExitQuestionnaire
"""
if assembly is None:
raise MigrationError(
"Parameter <assembly> is required to migrate exit questionnaire to version 6"
)
migrated_instance = self.convert_class(
self.new_model.RareDiseaseExitQuestionnaire, old_instance)
migrated_instance.variantGroupLevelQuestions = self.convert_collection(
old_instance.variantGroupLevelQuestions,
self._migrate_variant_group_level_question,
assembly=assembly)
return self.validate_object(
object_to_validate=migrated_instance,
object_type=self.new_model.RareDiseaseExitQuestionnaire)
def migrate_clinical_report_rd(self, old_instance, assembly):
"""
:type old_instance: reports_4_0_0.ClinicalReportRD
:type assembly: reports_5_0_0.Assembly
:rtype: reports_5_0_0.ClinicalReportRD
"""
if assembly is None:
raise MigrationError(
"Parameter <assembly> is required to migrate model versions earlier than 5.0.0"
)
new_instance = self.convert_class(
self.new_model.ClinicalReportRD,
old_instance) # :type self.new_model.ClinicalReportRD
try:
new_instance.interpretationRequestVersion = self.convert_string_to_integer(
old_instance.interpretationRequestVersion)
except MigrationError as ex:
logging.error(
"Error converting 'interpretationRequestVersion' to integer from value '{}'"
.format(old_instance.interpretationRequestVersion))
raise ex
new_instance.references = old_instance.supportingEvidence
new_instance.variants = self.convert_collection(
old_instance.candidateVariants,
self._migrate_reported_variant,
assembly=assembly)
if old_instance.additionalAnalysisPanels is not None:
panels = []
for panel in old_instance.additionalAnalysisPanels:
new_panel = self.new_model.AdditionalAnalysisPanel(
) # :type reports_5_0_0.AdditionalAnalysisPanel
new_panel.specificDisease = panel.specificDisease
new_panel.panel = self.new_model.GenePanel(
panelName=panel.panelName, panelVersion=panel.panelVersion)
panels.append(new_panel)
new_instance.additionalAnalysisPanels = panels
return self.validate_object(
object_to_validate=new_instance,
object_type=self.new_model.ClinicalReportRD)
def _merge_annotations_and_frequencies(numeric_annotations,
allele_frequencies):
if numeric_annotations is None:
numeric_annotations = {}
if not isinstance(numeric_annotations, dict):
raise MigrationError(
"additionalNumericVariantAnnotations should be dict but is: {}"
.format(numeric_annotations))
if allele_frequencies is not None:
for af in allele_frequencies:
annotation_key = "{}:{}".format(af.study, af.population)
if annotation_key in numeric_annotations:
logging.warning(
"{} already exists in numeric_annotations with value {} instead of {}"
.format(annotation_key,
numeric_annotations.get(annotation_key),
af.alternateFrequency))
else:
numeric_annotations["{}:{}".format(
af.study, af.population)] = af.alternateFrequency
return numeric_annotations
def migrate_interpretation_request_rd(self, old_instance, assembly):
"""
Migrates an InterpretationRequestRD into an InterpretedGenomeRD, several unexisting fields need to be provided
:type old_instance: reports_4_0_0.InterpretationRequestRD
:rtype: reports_5_0_0.InterpretationRequestRD
"""
if assembly is None:
raise MigrationError(
"Parameter <assembly> is required if version is older than 5.0.0"
)
new_instance = self.convert_class(
self.new_model.InterpretationRequestRD,
old_instance) # type: reports_5_0_0.InterpretationRequestRD
new_instance.genomeAssembly = assembly
new_instance.pedigree = self._migrate_pedigree(old_instance.pedigree)
# NOTE: store fields in additional fields that are lost otherwise
if not new_instance.additionalInfo:
new_instance.additionalInfo = {}
if old_instance.analysisVersion:
new_instance.additionalInfo[
'analysisVersion'] = old_instance.analysisVersion
if old_instance.analysisReturnUri:
new_instance.additionalInfo[
'analysisReturnUri'] = old_instance.analysisReturnUri
if old_instance.tieringVersion:
new_instance.additionalInfo[
'tieringVersion'] = old_instance.tieringVersion
if old_instance.complexGeneticPhenomena:
new_instance.additionalInfo['complexGeneticPhenomena'] = str(
old_instance.complexGeneticPhenomena)
if old_instance.cellbaseVersion:
new_instance.additionalInfo[
'cellbaseVersion'] = old_instance.cellbaseVersion
if old_instance.interpretGenome:
new_instance.additionalInfo['interpretGenome'] = str(
old_instance.interpretGenome)
return self.validate_object(
object_to_validate=new_instance,
object_type=self.new_model.InterpretationRequestRD)
def migrate(json_dict, types, migrations):
valid_types = MigrationHelpers.is_valid(json_dict, types)
if len(valid_types) == 0:
raise MigrationError("JSON dict not valid according to any model")
elif len(valid_types) == 1:
typ = valid_types[0]
elif len(valid_types) > 1:
observed_version = MigrationHelpers.get_version_control(json_dict)
typ = None
for t in valid_types:
expected_version = MigrationHelpers.get_version_control(t().toJsonDict())
if expected_version == observed_version:
# chooses the type having the correct expected version
typ = t
break
if typ is None:
# if none matched chooses the most conservative one
typ = valid_types[-1]
migrations_to_apply = migrations[0:types.index(typ) + 1]
migrated = typ.fromJsonDict(json_dict)
for migration in reversed(migrations_to_apply):
migrated = migration(migrated)
return migrated
def migrate_cancer_clinical_report(self, old_instance, assembly,
participant_id, sample_ids):
"""
NOTE: we migrate from a model where only one sample and one participant is supported, thus we do not need
a list of samples or participants
:type old_instance: reports_4_0_0.ClinicalReportCancer
:type assembly: reports_5_0_0.Assembly
:type participant_id: str
:type sample_ids: map[str (alleleOrigin)]: str - {'germline_variant': 'LP...', 'somatic_variant': 'LP...'}
:rtype: reports_5_0_0.ClinicalReportCancer
"""
if not sample_ids or not assembly or not participant_id:
raise MigrationError(
"Missing required fields to migrate cancer clinical report from 4.0.0 to 5.0.0"
)
new_instance = self.convert_class(
self.new_model.ClinicalReportCancer,
old_instance) # :type: reports_5_0_0.ClinicalReportCancer
try:
new_instance.interpretationRequestVersion = self.convert_string_to_integer(
old_instance.interpretationRequestVersion)
except MigrationError as ex:
logging.error(
"Error converting 'interpretationRequestVersion' to integer from value '{}'"
.format(old_instance.interpretationRequestVersion))
raise ex
new_instance.variants = self.convert_collection(
old_instance.candidateVariants,
self._migrate_reported_variant_cancer,
assembly=assembly,
participant_id=participant_id,
sample_ids=sample_ids)
return self.validate_object(
object_to_validate=new_instance,
object_type=self.new_model.ClinicalReportCancer)
def migrate_cancer_interpretation_request(self, old_instance, assembly):
"""
:type old_instance: reports_4_0_0.CancerInterpretationRequest
:rtype: reports_5_0_0.CancerInterpretationRequest
"""
if assembly is None:
raise MigrationError(
"Parameter <assembly> is required to migrate cancer interpretation request to version 5"
)
new_instance = self.convert_class(
self.new_model.CancerInterpretationRequest,
old_instance) # :type: reports_5_0_0.CancerInterpretationRequest
new_instance.interpretationRequestId = old_instance.reportRequestId
new_instance.interpretationRequestVersion = old_instance.reportVersion
new_instance.genomeAssembly = assembly
new_instance.cancerParticipant = self._migrate_cancer_participant(
old_participant=old_instance.cancerParticipant)
if not new_instance.additionalInfo:
new_instance.additionalInfo = {}
if old_instance.analysisUri:
new_instance.additionalInfo[
'analysisUri'] = old_instance.analysisUri
if old_instance.analysisVersion:
new_instance.additionalInfo[
'analysisVersion'] = old_instance.analysisVersion
if old_instance.tieringVersion:
new_instance.additionalInfo[
'tieringVersion'] = old_instance.tieringVersion
if old_instance.interpretGenome:
new_instance.additionalInfo['interpretGenome'] = str(
old_instance.interpretGenome)
return self.validate_object(
object_to_validate=new_instance,
object_type=self.new_model.CancerInterpretationRequest)
def _raise_migration_error_for_parameter(parameter):
raise MigrationError(
"Missing required field {parameter} to migrate a cancer interpreted genome from 4.0.0 to 5.0.0"
.format(parameter=parameter))