def toProtocolElement(self, tier=0):
"""
Returns the GA4GH protocol representation of this ReadGroup.
"""
# TODO this is very incomplete, but we don't have the
# implementation to fill out the rest of the fields currently
readGroup = protocol.ReadGroup()
readGroup.id = self.getId()
readGroup.created = self._creationTime
readGroup.updated = self._updateTime
dataset = self.getParentContainer().getParentContainer()
readGroup.dataset_id = dataset.getId()
readGroup.name = self.getLocalId()
readGroup.predicted_insert_size = pb.int(self.getPredictedInsertSize())
referenceSet = self._parentContainer.getReferenceSet()
readGroup.sample_name = pb.string(self.getSampleName())
readGroup.biosample_id = pb.string(self.getBiosampleId())
if referenceSet is not None:
readGroup.reference_set_id = referenceSet.getId()
readGroup.stats.CopyFrom(self.getStats())
readGroup.programs.extend(self.getPrograms())
readGroup.description = pb.string(self.getDescription())
readGroup.experiment.CopyFrom(self.getExperiment())
self.serializeAttributes(readGroup)
return readGroup
def _gaFeatureForFeatureDbRecord(self, feature):
"""
:param feature: The DB Row representing a feature
:return: the corresponding GA4GH protocol.Feature object
"""
gaFeature = protocol.Feature()
gaFeature.id = self.getCompoundIdForFeatureId(feature['id'])
if feature.get('parent_id'):
gaFeature.parent_id = self.getCompoundIdForFeatureId(
feature['parent_id'])
else:
gaFeature.parent_id = ""
gaFeature.feature_set_id = self.getId()
gaFeature.reference_name = pb.string(feature.get('reference_name'))
gaFeature.start = pb.int(feature.get('start'))
gaFeature.end = pb.int(feature.get('end'))
gaFeature.name = pb.string(feature.get('name'))
if feature.get('strand', '') == '-':
gaFeature.strand = protocol.NEG_STRAND
else:
# default to positive strand
gaFeature.strand = protocol.POS_STRAND
gaFeature.child_ids.extend(list(map(
self.getCompoundIdForFeatureId,
json.loads(feature['child_ids']))))
gaFeature.feature_type.CopyFrom(
self._ontology.getGaTermByName(feature['type']))
attributes = json.loads(feature['attributes'])
# TODO: Identify which values are ExternalIdentifiers and OntologyTerms
for key in attributes:
for v in attributes[key]:
gaFeature.attributes.attr[key].values.add().string_value = v
if 'gene_name' in attributes and len(attributes['gene_name']) > 0:
gaFeature.gene_symbol = pb.string(attributes['gene_name'][0])
return gaFeature
def toProtocolElement(self, tier=0):
dataset = protocol.Dataset()
dataset.id = self.getId()
dataset.name = pb.string(self.getLocalId())
dataset.description = pb.string(self.getDescription())
# Populate DUO info by extending the list
dataset.terms_of_use.extend([
protocol.fromJson(json.dumps(p), protocol.OntologyTerm)
for p in self.getInfo()
])
self.serializeAttributes(dataset)
return dataset
def toProtocolElement(self, tier=0):
"""
Returns the GA4GH protocol representation of this ReferenceSet.
"""
ret = protocol.ReferenceSet()
ret.assembly_id = pb.string(self.getAssemblyId())
ret.description = pb.string(self.getDescription())
ret.id = self.getId()
ret.is_derived = self.getIsDerived()
ret.md5checksum = self.getMd5Checksum()
if self.getSpecies():
term = protocol.fromJson(
json.dumps(self.getSpecies()), protocol.OntologyTerm)
ret.species.term_id = term.term_id
ret.species.term = term.term
ret.source_accessions.extend(self.getSourceAccessions())
ret.source_uri = pb.string(self.getSourceUri())
ret.name = self.getLocalId()
self.serializeAttributes(ret)
return ret
def toProtocolElement(self, tier=0):
"""
Returns the representation of this FeatureSet as the corresponding
ProtocolElement.
"""
gaFeatureSet = protocol.FeatureSet()
gaFeatureSet.id = self.getId()
gaFeatureSet.dataset_id = self.getParentContainer().getId()
gaFeatureSet.reference_set_id = pb.string(self._referenceSet.getId())
gaFeatureSet.name = self._name
gaFeatureSet.source_uri = self._sourceUri
attributes = self.getAttributes()
for key in attributes:
gaFeatureSet.attributes.attr[key] \
.values.extend(protocol.encodeValue(attributes[key]))
return gaFeatureSet
def getExperiment(self):
"""
Returns the GA4GH protocol representation of this read group's
Experiment.
"""
experiment = protocol.Experiment()
experiment.id = self.getExperimentId()
experiment.instrument_model = pb.string(self.getInstrumentModel())
experiment.sequencing_center = pb.string(self.getSequencingCenter())
experiment.description = pb.string(self.getExperimentDescription())
experiment.library = pb.string(self.getLibrary())
experiment.platform_unit = pb.string(self.getPlatformUnit())
experiment.message_create_time = self._iso8601
experiment.message_update_time = self._iso8601
experiment.run_time = pb.string(self.getRunTime())
return experiment
def testString(self):
self.assertEqual(pb.DEFAULT_STRING, pb.string(None))
self.assertEqual('A', pb.string('A'))