def __init__(
self, localId, randomSeed=1, numCalls=1,
variantDensity=1, numVariantSets=1, numAlignments=1):
super(SimulatedDataset, self).__init__(localId)
self._randomSeed = randomSeed
self._randomGenerator = random.Random()
self._randomGenerator.seed(self._randomSeed)
# Variants
for i in range(numVariantSets):
localId = "simVs{}".format(i)
seed = self._randomGenerator.randint(0, 2**32 - 1)
variantSet = variants.SimulatedVariantSet(
self, localId, seed, numCalls, variantDensity)
self._variantSetIdMap[variantSet.getId()] = variantSet
self._variantSetIds = sorted(self._variantSetIdMap.keys())
# Reads
localId = 'aReadGroupSet'
readGroupSet = reads.SimulatedReadGroupSet(
self, localId, numAlignments)
self._readGroupSetIdMap[readGroupSet.getId()] = readGroupSet
for readGroup in readGroupSet.getReadGroups():
self._readGroupIdMap[readGroup.getId()] = readGroup
self._readGroupSetIds = sorted(self._readGroupSetIdMap.keys())
self._readGroupIds = sorted(self._readGroupIdMap.keys())
def __init__(
self, datasetId, randomSeed, numCalls,
variantDensity, numVariantSets, numAlignments):
super(SimulatedDataset, self).__init__()
self._id = datasetId
self._randomSeed = randomSeed
self._randomGenerator = random.Random()
self._randomGenerator.seed(self._randomSeed)
# Variants
for i in range(numVariantSets):
variantSetId = "{}:simVs{}".format(self._id, i)
seed = self._randomGenerator.randint(0, 2**32 - 1)
variantSet = variants.SimulatedVariantSet(
seed, numCalls, variantDensity, variantSetId)
self._variantSetIdMap[variantSetId] = variantSet
self._variantSetIds = sorted(self._variantSetIdMap.keys())
# Reads
readGroupSetId = "{}:aReadGroupSet".format(self._id)
readGroupSet = reads.SimulatedReadGroupSet(
readGroupSetId, numAlignments)
self._readGroupSetIdMap[readGroupSetId] = readGroupSet
for readGroup in readGroupSet.getReadGroups():
self._readGroupIdMap[readGroup.getId()] = readGroup
self._readGroupSetIds = sorted(self._readGroupSetIdMap.keys())
self._readGroupIds = sorted(self._readGroupIdMap.keys())
def __init__(self,
localId,
referenceSet,
randomSeed=0,
numVariantSets=1,
numCalls=1,
variantDensity=0.5,
numReadGroupSets=1,
numReadGroupsPerReadGroupSet=1,
numAlignments=1,
numFeatureSets=1):
super(SimulatedDataset, self).__init__(localId)
self._description = "Simulated dataset {}".format(localId)
# TODO create a simulated Ontology
# Variants
for i in range(numVariantSets):
localId = "simVs{}".format(i)
seed = randomSeed + i
variantSet = variants.SimulatedVariantSet(self, referenceSet,
localId, seed, numCalls,
variantDensity)
callSets = variantSet.getCallSets()
# Add biosamples
for callSet in callSets:
bioSample = biodata.BioSample(self, callSet.getLocalId())
bioSample2 = biodata.BioSample(self,
callSet.getLocalId() + "2")
individual = biodata.Individual(self, callSet.getLocalId())
bioSample.setIndividualId(individual.getId())
bioSample2.setIndividualId(individual.getId())
self.addIndividual(individual)
self.addBioSample(bioSample)
self.addBioSample(bioSample2)
self.addVariantSet(variantSet)
variantAnnotationSet = variants.SimulatedVariantAnnotationSet(
variantSet, "simVas{}".format(i), seed)
variantSet.addVariantAnnotationSet(variantAnnotationSet)
# Reads
for i in range(numReadGroupSets):
localId = 'simRgs{}'.format(i)
seed = randomSeed + i
readGroupSet = reads.SimulatedReadGroupSet(
self, localId, referenceSet, seed,
numReadGroupsPerReadGroupSet, numAlignments)
for rg in readGroupSet.getReadGroups():
bioSample = biodata.BioSample(self, rg.getLocalId())
individual = biodata.Individual(self, rg.getLocalId())
bioSample.setIndividualId(individual.getId())
rg.setBioSampleId(bioSample.getId())
self.addIndividual(individual)
self.addBioSample(bioSample)
self.addReadGroupSet(readGroupSet)
# Features
for i in range(numFeatureSets):
localId = "simFs{}".format(i)
seed = randomSeed + i
featureSet = sequenceAnnotations.SimulatedFeatureSet(
self, localId, seed)
featureSet.setReferenceSet(referenceSet)
self.addFeatureSet(featureSet)
def __init__(self,
localId,
referenceSet,
randomSeed=0,
numVariantSets=1,
numCalls=1,
variantDensity=0.5,
numReadGroupSets=1,
numReadGroupsPerReadGroupSet=1,
numAlignments=1):
super(SimulatedDataset, self).__init__(localId)
self._description = "Simulated dataset {}".format(localId)
# Variants
for i in range(numVariantSets):
localId = "simVs{}".format(i)
seed = randomSeed + i
variantSet = variants.SimulatedVariantSet(self, localId, seed,
numCalls, variantDensity)
self.addVariantSet(variantSet)
# Reads
for i in range(numReadGroupSets):
localId = 'simRgs{}'.format(i)
seed = randomSeed + i
readGroupSet = reads.SimulatedReadGroupSet(
self, localId, referenceSet, seed,
numReadGroupsPerReadGroupSet, numAlignments)
self.addReadGroupSet(readGroupSet)
def testCreation(self):
dataset = datasets.AbstractDataset('dataset1')
localId = "readGroupSetId"
simulatedReadGroupSet = reads.SimulatedReadGroupSet(
dataset, localId)
for readGroup in simulatedReadGroupSet.getReadGroups():
alignments = list(readGroup.getReadAlignments())
self.assertGreater(len(alignments), 0)
def testCreation(self):
dataset = datasets.Dataset('dataset1')
localId = "readGroupSetId"
referenceSet = references.SimulatedReferenceSet("srs1")
simulatedReadGroupSet = reads.SimulatedReadGroupSet(
dataset, localId, referenceSet)
for readGroup in simulatedReadGroupSet.getReadGroups():
alignments = list(readGroup.getReadAlignments())
self.assertGreater(len(alignments), 0)
def __init__(self, randomSeed=0, numCalls=1, variantDensity=0.5,
numVariantSets=1):
super(SimulatedBackend, self).__init__()
self._randomSeed = randomSeed
self._randomGenerator = random.Random()
self._randomGenerator.seed(self._randomSeed)
for i in range(numVariantSets):
variantSetId = "simVs{}".format(i)
seed = self._randomGenerator.randint(0, 2**32 - 1)
variantSet = variants.SimulatedVariantSet(
seed, numCalls, variantDensity, variantSetId)
self._variantSetIdMap[variantSetId] = variantSet
self._variantSetIds = sorted(self._variantSetIdMap.keys())
# Reads
readGroupSetId = "aReadGroupSet"
readGroupSet = reads.SimulatedReadGroupSet(readGroupSetId)
self._readGroupSetIdMap[readGroupSetId] = readGroupSet
for readGroup in readGroupSet.getReadGroups():
self._readGroupIdMap[readGroup.getId()] = readGroup
self._readGroupSetIds = sorted(self._readGroupSetIdMap.keys())
self._readGroupIds = sorted(self._readGroupIdMap.keys())
def testCreation(self):
readGroupSetId = "readGroupSetId"
simulatedReadGroupSet = reads.SimulatedReadGroupSet(readGroupSetId)
for readGroup in simulatedReadGroupSet.getReadGroups():
alignments = list(readGroup.getReadAlignments())
self.assertGreater(len(alignments), 0)