def test_add_new_call_set(self):
vcf = VCF(f="tests/vcf_tests/test.vcf",
reference_set_id=self.reference_set.id,
method="CORTEX")
vcf.add_to_database()
# Only one callset but the callset should belong to multiple variant
# sets
assert VariantCallSet.objects().count() == 1
assert VariantCallSet.objects()[0].created_at <= datetime.datetime.now(
)
assert len(VariantCallSet.objects()[0].variant_sets) == 2
def setUp(self):
self.reference_set = ReferenceSet().create_and_save(name="ref_set")
self.variant_set = VariantSet.create_and_save(
name="this_vcf_file2",
reference_set=self.reference_set)
self.variant_sets = [self.variant_set]
self.reference = Reference().create_and_save(
name="ref",
md5checksum="sre",
reference_sets=[
self.reference_set])
self.call_set = VariantCallSet.create(
sample_id="C00123",
name="C00123",
variant_sets=self.variant_sets)
self.variant_snp = Variant.create(variant_sets=self.variant_sets,
start=0, end=1, reference_bases="A",
alternate_bases=["T"],
reference=self.reference)
self.variant_snp_mult_alts = Variant.create(
variant_sets=self.variant_sets,
start=0,
end=1,
reference_bases="T",
alternate_bases=[
"A",
"C"],
reference=self.reference)
def setUp(self):
self.reference_set = ReferenceSet().create_and_save(name="ref_set")
self.variant_set = VariantSet.create_and_save(
name="this_vcf_file2",
reference_set=self.reference_set)
self.variant_sets = [self.variant_set]
self.reference = Reference().create_and_save(
name="ref",
md5checksum="sre",
reference_sets=[
self.reference_set])
self.call_set = VariantCallSet.create(
sample_id="C00123",
name="C00123",
variant_sets=self.variant_sets)
self.variant_snp = Variant.create(variant_sets=self.variant_sets,
start=0, end=1, reference_bases="A",
alternate_bases=["T"],
reference=self.reference)
self.variant_snp_mult_alts = Variant.create(
variant_sets=self.variant_sets,
start=0,
end=1,
reference_bases="T",
alternate_bases=[
"A",
"C"],
reference=self.reference)
def test_create_call_set(self):
call_set = VariantCallSet.create_and_save(
name="call_set",
sample_id="C00123",
variant_sets=self.variant_sets)
cs = VariantCallSet.objects.get(name="call_set")
assert call_set == cs
assert cs.name == "call_set"
assert cs.variant_sets[0].reference_set.name == "ref_set"
def test_create_call_set(self):
call_set = VariantCallSet.create_and_save(
name="call_set",
sample_id="C00123",
variant_sets=self.variant_sets)
cs = VariantCallSet.objects.get(name="call_set")
assert call_set == cs
assert cs.name == "call_set"
assert cs.variant_sets[0].reference_set.name == "ref_set"
def _remove_variant_set(self, variant_set_name):
vs = VariantSet.objects.get(name=variant_set_name,
reference_set=self.reference_set)
for call_set in VariantCallSet.objects(variant_sets=vs):
call_set.variant_sets.remove(vs)
call_set.save()
# Remove calls from callsets that only have this variantset
if len(call_set.variant_sets) < 2:
VariantCall.objects(call_set=call_set).delete()
call_set.delete()
# Remove variants that are ONLY from this variant set
Variant.objects(variant_sets=vs, variant_sets__size=2).delete()
VariantSetMetadata.objects(variant_set=vs).delete()
vs.delete()
def _create_call_sets(self):
for sample in self.vcf_reader.samples:
try:
cs = VariantCallSet.create_and_save(
name="_".join([sample, self.method]),
variant_sets=self.variant_sets,
sample_id=sample,
info={"variant_caller": self.method})
except NotUniqueError:
raise ValueError(
"There is already a call set for sample %s with method %s "
% (sample, self.method))
else:
self.call_sets[sample] = cs
def test_add_second_vcf_variant_set(self):
# This VCF only has one Variant which is not in the first VCF
vcf = VCF(f="tests/vcf_tests/test2.vcf",
reference_set_id=self.reference_set.id,
method="CORTEX")
vcf.add_to_database()
assert VariantSet.objects().count() == 3
assert VariantCallSet.objects().count() == 2
assert VariantCall.objects().count() == 42
assert Variant.objects().count() == 22
assert len(Variant.objects()[0].variant_sets) == 3
assert len(
Variant.objects.get(
names="UNION_BC_k31_var_147").variant_sets) == 3
def test_add_second_vcf_variant_set(self):
# This VCF only has one Variant which is not in the first VCF
vcf = VCF(f="tests/vcf_tests/test3.vcf",
reference_set_id=self.reference_set.id,
method="CORTEX")
vcf.add_to_database()
assert VariantSet.objects().count() == 2
assert VariantCallSet.objects().count() == 1
assert VariantCall.objects().count() == 106
assert Variant.objects().count() == 106
assert Variant.snps().count() == 89
assert Variant.indels().count() == 17
assert Variant.insertions().count() == 8
assert Variant.deletions().count() == 8
assert Variant.ph_snps.count() == 1
def _create_call_sets(self):
try:
self.call_set = VariantCallSet.create_and_save(
name=self.call_set_name,
variant_sets=[self.global_variant_set],
sample_id=self.sample,
info={"variant_caller": self.method})
except NotUniqueError:
if self.force:
# logger.info("There is already a call set for sample %s with method %s but add has been called with --force so this callset is being removed from the DB" %
# (self.sample, self.method))
cs = VariantCallSet.objects.get(
name="_".join([self.sample, self.method]))
VariantCall.objects(call_set=cs).delete()
cs.delete()
self._create_call_sets()
else:
raise ValueError(
"There is already a call set for sample %s with method %s "
% (self.sample, self.method))