def setUp(self):
""" define a default Person object
"""
ID = "fam_ID"
self.family = Family(ID)
def test_load_trio(self):
''' test that load_trio() works correctly
'''
def make_vcf(person):
# make a VCF, where one line would pass the default filtering
vcf = make_vcf_header()
vcf.append(make_vcf_line(pos=1, extra='HGNC=TEST;MAX_AF=0.0001'))
vcf.append(make_vcf_line(pos=2, extra='HGNC=ATRX;MAX_AF=0.0001'))
path = os.path.join(self.temp_dir, "{}.vcf.gz".format(person))
self.write_gzipped_vcf(path, vcf)
return path
child_path = make_vcf('child')
mother_path = make_vcf('mother')
father_path = make_vcf('father')
family = Family('fam_id')
family.add_child('sample', 'mother_id', 'father_id', 'female', '2', child_path)
family.add_mother('mother_id', '0', '0', 'female', '1', mother_path)
family.add_father('father_id', '0', '0', 'male', '1', father_path)
family.set_child()
# define the parameters and values for the SNV class
args = {'chrom': "1", 'position': 2, 'id': ".", 'ref': "G", 'alts': "T",
'filter': "PASS", 'info': "CQ=missense_variant;HGNC=ATRX;MAX_AF=0.0001",
'format': "DP:GT", 'sample': "50:0/1", 'gender': "female",
'mnv_code': None}
dad_args = copy.deepcopy(args)
dad_args['gender'] = 'male'
self.assertEqual(self.vcf_loader.load_trio(family),
[TrioGenotypes(chrom="1", pos=2, child=SNV(**args),
mother=SNV(**args), father=SNV(**dad_args)) ])
def create_family(self, child_gender, mom_aff, dad_aff):
""" create a default family, with optional gender and parental statuses
"""
fam = Family("test")
fam.add_child("child", "child_vcf", "2", child_gender)
fam.add_mother("mother", "mother_vcf", mom_aff, "2")
fam.add_father("father", "father_vcf", dad_aff, "1")
fam.set_child()
return fam
def create_family(self, child_gender, mom_aff, dad_aff):
""" create a default family, with optional gender and parental statuses
"""
fam = Family('test')
fam.add_child('child', 'mother', 'father', child_gender, '2',
'child_vcf')
fam.add_mother('mother', '0', '0', 'female', mom_aff, 'mother_vcf')
fam.add_father('father', '0', '0', 'male', dad_aff, 'father_vcf')
fam.set_child()
return fam
def test_load_trio(self):
''' test that load_trio() works correctly
'''
def make_vcf(person):
# make a VCF, where one line would pass the default filtering
vcf = make_vcf_header()
vcf.append(make_vcf_line(pos=1, extra='HGNC=TEST;MAX_AF=0.0001'))
vcf.append(make_vcf_line(pos=2, extra='HGNC=ATRX;MAX_AF=0.0001'))
path = os.path.join(self.temp_dir, "{}.vcf.gz".format(person))
write_gzipped_vcf(path, vcf)
return path
child_path = make_vcf('child')
mother_path = make_vcf('mother')
father_path = make_vcf('father')
family = Family('fam_id')
family.add_child('sample', 'mother_id', 'father_id', 'female', '2',
child_path)
family.add_mother('mother_id', '0', '0', 'female', '1', mother_path)
family.add_father('father_id', '0', '0', 'male', '1', father_path)
family.set_child()
sum_x_lr2_proband = 0
# define the parameters and values for the SNV class
args = {
'chrom': "1",
'position': 2,
'id': ".",
'ref': "G",
'alts': "T",
'filter': "PASS",
'info': "CQ=missense_variant;HGNC=ATRX;MAX_AF=0.0001",
'format': "DP:GT:AD",
'sample': "50:0/1:10,10",
'gender': "female",
'mnv_code': None,
'qual': '1000'
}
dad_args = copy.deepcopy(args)
dad_args['gender'] = 'male'
self.assertEqual(load_trio(family, sum_x_lr2_proband), [
TrioGenotypes(chrom="1",
pos=2,
child=SNV(**args),
mother=SNV(**args),
father=SNV(**dad_args))
])
def test_get_vcf_provenance(self):
""" test that get_vcf_provenance() works correctly
"""
path = os.path.join(self.temp_dir, "temp.vcf")
gz_path = os.path.join(self.temp_dir, "temp.vcf.gz")
date_path = os.path.join(self.temp_dir, "temp.process.2014-02-20.vcf")
family = Family('famid')
family.add_child('child_id', 'mother', 'father', 'f', '2', path)
family.add_mother('mom_id', '0', '0', 'female', '1', gz_path)
family.add_father('dad_id', '0', '0', 'male', '1', date_path)
family.set_child()
vcf = make_minimal_vcf()
vcf_string = "".join(vcf)
if IS_PYTHON3:
vcf_string = vcf_string.encode("utf-8")
ungzipped_hash = hashlib.sha1(vcf_string).hexdigest()
header = vcf[:4]
write_temp_vcf(path, vcf)
# check that the file defs return correctly
(checksum, basename, date) = get_vcf_provenance(family.child)
self.assertEqual(checksum, ungzipped_hash)
self.assertEqual(basename, "temp.vcf")
self.assertEqual(date, "2014-01-01")
# now write a gzip file, and check that we get the correct hash
write_gzipped_vcf(gz_path, vcf)
handle = open(gz_path, "rb")
gzipped_hash = hashlib.sha1(handle.read()).hexdigest()
handle.close()
(checksum, basename, date) = get_vcf_provenance(family.mother)
self.assertEqual(checksum, gzipped_hash)
# check that when a fileDate isn't available in the VCF, we can pick
# the date from the path
vcf.pop(1)
write_temp_vcf(date_path, vcf)
(checksum, basename, date) = get_vcf_provenance(family.father)
self.assertEqual(date, "2014-02-20")
# and check we get null values if the family member is not present
family.father = None
provenance = get_vcf_provenance(family.father)
self.assertEqual(provenance, ('NA', 'NA', 'NA'))
def create_family(self, child_gender, mom_aff, dad_aff):
""" create a default family, with optional gender and parental statuses
"""
fam = Family("test")
fam.add_child("child", "child_vcf", "2", child_gender)
fam.add_mother("mother", "mother_vcf", mom_aff, "2")
fam.add_father("father", "father_vcf", dad_aff, "1")
fam.set_child()
return fam
def create_family(self, child_gender, mom_aff, dad_aff):
""" create a default family, with optional gender and parental statuses
"""
fam = Family('test')
fam.add_child('child', 'mother', 'father', child_gender, '2', 'child_vcf')
fam.add_mother('mother', '0', '0', 'female', mom_aff, 'mother_vcf')
fam.add_father('father', '0', '0', 'male', dad_aff, 'father_vcf')
fam.set_child()
return fam
def test_debug_option(self):
""" test whether we can set up the class with the debug option
"""
known = {}
pops = None
sum_x_lr2 = {}
vcf = make_minimal_vcf()
path = os.path.join(self.temp_dir, "temp.vcf.gz")
write_gzipped_vcf(path, vcf)
fam = Family(
'fam', children=[Person('fam', 'child', '0', '0', 'f', '2', path)])
# if the debug info isn't available, then the SNV object doesn't use the
# debug filter function
variants = load_variants(fam, 1.0, pops, known, set(), sum_x_lr2)
self.assertNotEqual(SNV.passes_filters, SNV.passes_filters_with_debug)
# if the debug info is passed in, check that the debug filter function
# got set correctly
variants = load_variants(fam, 1.0, pops, known, set(), sum_x_lr2, "1",
"10000")
self.assertEqual(SNV.passes_filters, SNV.passes_filters_with_debug)
def test_is_compound_pair_proband_only(self):
""" check that is_compound_pair() includes proband-only pairs
"""
fam = Family("test")
fam.add_child("child", 'dad_id', 'mom_id', 'F', '2', "child_vcf")
fam.set_child()
# set some variants, so we can alter them later
var1 = self.create_variant(chrom="1", position="150", sex="F", cq="stop_gained")
var2 = self.create_variant(chrom="1", position="160", sex="F", cq="stop_gained")
inh = Autosomal([var1, var2], fam, self.known_gene, "TEST")
# check that a proband-only passes, regardless of the parental genotypes
self.assertTrue(inh.is_compound_pair(var1, var2))
def test_load_variants(self):
''' test that load_variants() works correctly. Mainly checks variables
are set correctly.
'''
vcf = make_minimal_vcf()
path = os.path.join(self.temp_dir, "temp.vcf.gz")
write_gzipped_vcf(path, vcf)
sum_x_lr2 = {}
parents = True
fam = Family(
'fam', children=[Person('fam', 'child', '0', '0', 'f', '2', path)])
variants = load_variants(fam, 0.9, ['AFR_AF'], self.known_genes, set(),
sum_x_lr2, parents)
self.assertEqual(SNV.known_genes, self.known_genes)
self.assertEqual(CNV.known_genes, self.known_genes)
self.assertEqual(Info.populations, ['AFR_AF'])
self.assertEqual(Info.last_base, set())
# and check that the
variants = load_variants(fam, 0.9, [], None, set([('1', 100)]),
sum_x_lr2, parents)
self.assertIsNone(SNV.known_genes, self.known_genes)
self.assertIsNone(CNV.known_genes, self.known_genes)
self.assertEqual(Info.populations, [])
self.assertEqual(Info.last_base, set([('1', 100)]))
def setUp(self):
""" define a default Person object
"""
ID = "fam_ID"
self.family = Family(ID)
def test_find_variants(self):
""" test that find_variants() works correctly
"""
# define the trio, so that we can know whether the parents are affected.
# The child also needs to be included and set, so that we can get the
# child ID for logging purposes.
family = Family("famID")
family.add_child("child_id", 'dad_id', 'mom_id', 'f', '2', "/vcf/path")
family.add_father("dad_id", '0', '0', 'm', '1', "/vcf/path")
family.add_mother("mom_id", '0', '0', 'f', '1', "/vcf/path")
family.set_child()
# create variants that cover various scenarios
snv1 = create_variant("F", "missense_variant|missense_variant", "TEST1|TEST2")
snv2 = create_variant("F", "missense_variant|synonymous_variant", "OTHER1|OTHER2")
snv3 = create_variant("F", "missense_variant", "")
snv4 = create_variant("F", "missense_variant", "TESTX", chrom="X")
self.finder.known_genes = {"TEST1": {"inh": ["Monoallelic"]},
"OTHER1": {"inh": ["Monoallelic"]},
"OTHER2": {"inh": ["Monoallelic"]},
"TESTX": {"inh": ["X-linked dominant"]}}
# check the simplest case, a variant in a known gene
self.assertEqual(self.finder.find_variants([snv1], "TEST1", family),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"])])
# check that a gene not in a known gene does not pass
self.assertEqual(self.finder.find_variants([snv1], "TEST2", family), [])
# check a variant where the gene is known, but the consequence for that
# gene is not functional, does not pass
self.assertEqual(self.finder.find_variants([snv2], "OTHER2", family), [])
# check that intergenic variants (which lack HGNC symbols) do not pass
self.assertEqual(self.finder.find_variants([snv3], None, family), [])
# check that a variant on chrX passes through the allosomal instance
self.assertEqual(self.finder.find_variants([snv4], "TESTX", family),
[(snv4, ["single_variant"], ["X-linked dominant"], ["TESTX"])])
# remove the known genes, so that the variants in unknown genes pass
self.finder.known_genes = None
self.assertEqual(sorted(self.finder.find_variants([snv1], "TEST2", family)),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST2"]),
(snv1, ["single_variant"], ["Mosaic"], ["TEST2"])])
# but variants without gene symbols still are excluded
self.assertEqual(self.finder.find_variants([snv3], None, family), [])
def setUp(self):
""" define a default VcfInfo object
"""
family = Family("FamID")
family.add_child("child_id", "/child/path", "2", "M")
family.add_mother("mom_id", "/mother/path", "1", "F")
family.add_father("dad_id", "/father/path", "2", "M")
family.set_child()
variants = []
snv = self.create_var("1", True)
cnv = self.create_var("1", False)
variants.append((snv, ["single_variant"], ["Monoallelic"], ["ATRX"]))
variants.append((cnv, ["single_variant"], ["Monoallelic"], ["ATRX"]))
self.post_filter = PostInheritanceFilter(variants, family)
def test_filter_de_novos(self):
""" check that filter_de_novos() works correctly
"""
# make a family without parents
family = Family("fam_id")
child_gender = "female"
family.add_child("child_id", "child_vcf_path", "2", child_gender)
self.vcf_loader.family = family
# set up an autosomal variant
line = ["1", "100", ".", "T", "G", "1000", "PASS", ".", "GT", "0/1"]
gender = "M"
child_var = SNV(*line[:6])
child_var.add_info(line[7])
child_var.add_format(line[8], line[9])
child_var.set_gender(child_gender)
child_var.set_genotype()
# combine the variant into a list of TrioGenotypes
child_vars = [child_var]
mother_vars = []
father_vars = []
trio_variants = self.vcf_loader.combine_trio_variants(
child_vars, mother_vars, father_vars)
# check that vars without parents get passed through automatically
self.assertEqual(self.vcf_loader.filter_de_novos(trio_variants, 0.9),
trio_variants)
# now add parents to the family
family.add_mother("mother_id", "mother_vcf_path", "1", "female")
family.add_father("father_id", "father_vcf_path", "1", "male")
# re-generate the variants list now that parents have been included
trio_variants = self.vcf_loader.combine_trio_variants(
child_vars, mother_vars, father_vars)
# check that vars with parents, and that appear to be de novo are
# filtered out
self.assertEqual(self.vcf_loader.filter_de_novos(trio_variants, 0.9),
[])
# check that vars with parents, but which are not de novo, are retained
mother_vars = child_vars
trio_variants = self.vcf_loader.combine_trio_variants(
child_vars, mother_vars, father_vars)
self.assertEqual(self.vcf_loader.filter_de_novos(trio_variants, 0.9),
trio_variants)
def setUp(self):
""" define a default variant object
"""
Info.populations = ['AFR_AF']
family = Family('test')
family.add_child('child', 'mother', 'father', 'male', '2', 'child_vcf')
family.add_mother('mother', '0', '0', 'female', '1', 'mother_vcf')
family.add_father('father', '0', '0', 'male', '2', 'father_vcf')
family.set_child()
self.variants = []
snv = self.create_var("1", True)
cnv = self.create_var("1", False)
self.variants.append(
(snv, ["single_variant"], ["Monoallelic"], ["ATRX"]))
self.variants.append(
(cnv, ["single_variant"], ["Monoallelic"], ["ATRX"]))
self.post_filter = PostInheritanceFilter(family)
def test_filter_de_novos(self):
""" check that filter_de_novos() works correctly
"""
# make a family without parents
family = Family("fam_id")
child_gender = "female"
family.add_child('child_id', 'mother_id', 'father_id', child_gender,
'2', 'child_path')
# set up an autosomal variant
gender = "M"
args = [
"1", "100", ".", "T", "G", "1000", "PASS", ".", "GT", "0/1", gender
]
child_var = SNV(*args)
# combine the variant into a list of TrioGenotypes
child_vars = [child_var]
mother_vars = []
father_vars = []
trio_variants = combine_trio_variants(family, child_vars, mother_vars,
father_vars)
# check that vars without parents get passed through automatically
self.assertEqual(filter_de_novos(trio_variants, 0.9), trio_variants)
# now add parents to the family
family.add_mother("mother_id", '0', '0', 'female', '1',
"mother_vcf_path")
family.add_father("father_id", '0', '0', 'male', '1',
"father_vcf_path")
family = family
# re-generate the variants list now that parents have been included
trio_variants = combine_trio_variants(family, child_vars, mother_vars,
father_vars)
# check that vars with parents, and that appear to be de novo are
# filtered out
self.assertEqual(filter_de_novos(trio_variants, 0.9), [])
# check that vars with parents, but which are not de novo, are retained
mother_vars = child_vars
trio_variants = combine_trio_variants(family, child_vars, mother_vars,
father_vars)
self.assertEqual(filter_de_novos(trio_variants, 0.9), trio_variants)
def setUp(self):
""" define a default variant object
"""
Info.populations = ['AFR_AF']
family = Family('test')
family.add_child('child', 'mother', 'father', 'male', '2', 'child_vcf')
family.add_mother('mother', '0', '0', 'female', '1', 'mother_vcf')
family.add_father('father', '0', '0', 'male', '2', 'father_vcf')
family.set_child()
self.variants = []
snv = self.create_var("1", True)
cnv = self.create_var("1", False)
self.variants.append((snv, ["single_variant"], ["Monoallelic"], ["ATRX"]))
self.variants.append((cnv, ["single_variant"], ["Monoallelic"], ["ATRX"]))
self.post_filter = PostInheritanceFilter(family)
def get_families(args):
""" loads a list of Family objects for multiple families, or a single trio
"""
if args.ped is None:
fam_id = 'blank_family_ID'
family = Family(fam_id)
family.add_child('child', args.mother, args.father, args.gender, '2',
args.child)
if args.mother is not None:
family.add_mother('mother', '0', '0', '2', args.mom_aff,
args.mother)
if args.father is not None:
family.add_father('father', '0', '0', '1', args.dad_aff,
args.father)
families = [family]
else:
families = load_families(args.ped)
return families
def get_families(args):
""" loads a list of Family objects for multiple families, or a single trio
"""
if args.ped is None:
fam_id = 'blank_family_ID'
family = Family(fam_id)
family.add_child('child', args.mother, args.father, args.gender, '2', args.child)
if args.mother is not None:
family.add_mother('mother', '0', '0', '2', args.mom_aff, args.mother)
if args.father is not None:
family.add_father('father', '0', '0', '1', args.dad_aff, args.father)
families = [family]
else:
families = load_families(args.ped)
return families
def test_filter_de_novos(self):
""" check that filter_de_novos() works correctly
"""
# make a family without parents
family = Family("fam_id")
child_gender = "female"
family.add_child('child_id', 'mother_id', 'father_id', child_gender, '2', 'child_path')
self.vcf_loader.family = family
# set up an autosomal variant
gender = "M"
args = ["1", "100", ".", "T", "G", "PASS", ".", "GT", "0/1", gender]
child_var = SNV(*args)
# combine the variant into a list of TrioGenotypes
child_vars = [child_var]
mother_vars = []
father_vars = []
trio_variants = self.vcf_loader.combine_trio_variants(family, child_vars, mother_vars, father_vars)
# check that vars without parents get passed through automatically
self.assertEqual(self.vcf_loader.filter_de_novos(trio_variants, 0.9), trio_variants)
# now add parents to the family
family.add_mother("mother_id", '0', '0', 'female', '1', "mother_vcf_path")
family.add_father("father_id", '0', '0', 'male', '1', "father_vcf_path")
self.vcf_loader.family = family
# re-generate the variants list now that parents have been included
trio_variants = self.vcf_loader.combine_trio_variants(family, child_vars, mother_vars, father_vars)
# check that vars with parents, and that appear to be de novo are
# filtered out
self.assertEqual(self.vcf_loader.filter_de_novos(trio_variants, 0.9), [])
# check that vars with parents, but which are not de novo, are retained
mother_vars = child_vars
trio_variants = self.vcf_loader.combine_trio_variants(family, child_vars, mother_vars, father_vars)
self.assertEqual(self.vcf_loader.filter_de_novos(trio_variants, 0.9), trio_variants)
def test_is_compound_pair_proband_only(self):
""" check that is_compound_pair() includes proband-only pairs
"""
fam = Family("test")
fam.add_child("child", 'dad_id', 'mom_id', 'F', '2', "child_vcf")
fam.set_child()
# set some variants, so we can alter them later
var1 = self.create_variant(chrom="1",
position="150",
sex="F",
cq="stop_gained")
var2 = self.create_variant(chrom="1",
position="160",
sex="F",
cq="stop_gained")
inh = Autosomal([var1, var2], fam, self.known_gene, "TEST")
# check that a proband-only passes, regardless of the parental genotypes
self.assertTrue(inh.is_compound_pair(var1, var2))
def test_load_families(self):
""" check that load_families works correctly
"""
# construct a temporary family that will have the same sample IDs etc
# as for the one loaded from the ped file.
family = Family("fam_ID")
family.add_child("proband", 'dad', 'mom', 'F', '2',
"/path/to/proband_vcf.gz")
family.add_mother("mom", '0', '0', 'F', '1', "/path/to/mom_vcf.gz")
family.add_father("dad", '0', '0', 'M', '1', "/path/to/dad_vcf.gz")
# load the ped file, and check that the load_families function returns
# the expected Family object
self.assertEqual(load_families(self.path), [family])
# add an extra family, with multiple sibs
self.tempfile.write(
"fam_ID2 proband2 dad2 mom2 F 2 /path/to/proband2_vcf.gz\n")
self.tempfile.write(
"fam_ID2 dad2 0 0 M 1 /path/to/dad2_vcf.gz\n")
self.tempfile.write(
"fam_ID2 mom2 0 0 F 1 /path/to/mom2_vcf.gz\n")
self.tempfile.write(
"fam_ID2 sib dad2 mom2 F 2 /path/to/sib_vcf.gz\n")
self.tempfile.flush()
# construct a temporary family that will have the same sample IDs etc
# as for the one loaded from the ped file.
fam2 = Family("fam_ID2")
fam2.add_child("proband2", 'dad2', 'mom2', 'F', '2',
"/path/to/proband2_vcf.gz")
fam2.add_child("sib", 'dad2', 'mom2', 'F', '2', "/path/to/sib_vcf.gz")
fam2.add_mother("mom2", '0', '0', 'F', '1', "/path/to/mom2_vcf.gz")
fam2.add_father("dad2", '0', '0', 'M', '1', "/path/to/dad2_vcf.gz")
# load the ped file, and check that the load_families function returns
# the expected Families objects
self.assertEqual(sorted(load_families(self.path)),
sorted([family, fam2]))
def test_get_vcf_provenance(self):
""" test that get_vcf_provenance() works correctly
"""
path = os.path.join(self.temp_dir, "temp.vcf")
gz_path = os.path.join(self.temp_dir, "temp.vcf.gz")
date_path = os.path.join(self.temp_dir, "temp.process.2014-02-20.vcf")
family = Family('famid')
family.add_child('child_id', 'mother', 'father', 'f', '2', path)
family.add_mother('mom_id', '0', '0', 'female', '1', gz_path)
family.add_father('dad_id', '0', '0', 'male', '1', date_path)
family.set_child()
vcf = make_minimal_vcf()
vcf_string = "".join(vcf)
if IS_PYTHON3:
vcf_string = vcf_string.encode("utf-8")
ungzipped_hash = hashlib.sha1(vcf_string).hexdigest()
header = vcf[:4]
write_temp_vcf(path, vcf)
# check that the file defs return correctly
(checksum, basename, date) = get_vcf_provenance(family.child)
self.assertEqual(checksum, ungzipped_hash)
self.assertEqual(basename, "temp.vcf")
self.assertEqual(date, "2014-01-01")
# now write a gzip file, and check that we get the correct hash
write_gzipped_vcf(gz_path, vcf)
handle = open(gz_path, "rb")
gzipped_hash = hashlib.sha1(handle.read()).hexdigest()
handle.close()
(checksum, basename, date) = get_vcf_provenance(family.mother)
self.assertEqual(checksum, gzipped_hash)
# check that when a fileDate isn't available in the VCF, we can pick
# the date from the path
vcf.pop(1)
write_temp_vcf(date_path, vcf)
(checksum, basename, date) = get_vcf_provenance(family.father)
self.assertEqual(date, "2014-02-20")
# and check we get null values if the family member is not present
family.father = None
provenance = get_vcf_provenance(family.father)
self.assertEqual(provenance, ('NA', 'NA', 'NA'))
class TestFamily(unittest.TestCase):
"""
"""
def setUp(self):
""" define a default Person object
"""
ID = "fam_ID"
self.family = Family(ID)
def test__iter__(self):
''' test that __iter__() works correctly
'''
family_id, person_id, mom_id, dad_id = 'fam_ID', 'child', 'dad', 'mom',
path, status, sex = 'child.vcf', '2', 'M'
child = Person(family_id, person_id, dad_id, mom_id, sex, status, path)
self.family.add_child(person_id, dad_id, mom_id, sex, status, path)
self.family.set_child()
# check the Family iterates by getting a list of the Family object
members = list(self.family)
self.assertEqual(members, [child, None, None])
def test_add_father(self):
""" test that add_father() works correctly
"""
person_id = "parent_ID"
path = "/home/parent.vcf"
status = "1"
sex = "1"
dad_id = "0"
mom_id = "0"
# check that adding a male father doesn't raise an error
self.family.add_father(person_id, dad_id, mom_id, sex, status, path)
# check that adding a father for a second time is fine, but adding
# a different father raises an error
self.family.add_father(person_id, dad_id, mom_id, sex, status, path)
with self.assertRaises(ValueError):
self.family.add_father("different_ID", dad_id, mom_id, sex, status,
path)
# check that adding a female father raises an error
self.setUp()
sex = "2"
with self.assertRaises(ValueError):
self.family.add_father(person_id, dad_id, mom_id, sex, status,
path)
def test_add_mother(self):
""" test that add_mother() works correctly
"""
person_id = "parent_ID"
path = "/home/parent.vcf"
status = "1"
sex = "2"
dad_id = "0"
mom_id = "0"
# check that adding a female mother doesn't raise an error
self.family.add_mother(person_id, dad_id, mom_id, sex, status, path)
# check that adding a mother for a second time is fine, but adding
# a different mother raises an error
self.family.add_mother(person_id, dad_id, mom_id, sex, status, path)
with self.assertRaises(ValueError):
self.family.add_mother("different_ID", dad_id, mom_id, sex, status,
path)
# check that adding a male mother raises an error
self.setUp()
sex = "1"
with self.assertRaises(ValueError):
self.family.add_mother(person_id, dad_id, mom_id, sex, status,
path)
def test_add_child(self):
""" check that add_child() works correctly
"""
# check that we can add one child
self.family.add_child("child1", 'dad', 'mom', 'male', '2',
"/home/child1.vcf")
self.assertEqual(len(self.family.children), 1)
# check that adding multiple children works correctly
self.family.add_child("child2", 'dad', 'mom', 'female', '2',
"/home/child2.vcf")
self.family.add_child("child3", 'dad', 'mom', 'male', '2',
"/home/child3.vcf")
self.assertEqual(len(self.family.children), 3)
def test_set_child(self):
""" test that set_child() works correctly
"""
# add one child
self.family.add_child("child1", 'dad', 'mom', 'male', '2',
"/home/child1.vcf")
# check that the child can be set correctly
self.family.set_child()
self.assertEqual(self.family.child, self.family.children[0])
# add more children
self.family.add_child("child2", 'dad', 'mom', 'male', '2',
"/home/child2.vcf")
self.family.add_child("child3", 'dad', 'mom', 'female', '2',
"/home/child3.vcf")
# check that the child can be set correctly with multiple children
self.family.set_child()
self.assertIn(self.family.child, self.family.children)
def test_set_child_examined(self):
""" test that set_child_examined() works correctly
"""
# add one child
self.family.add_child("child1", 'dad', 'mom', 'male', '2',
"/home/child1.vcf")
# check that the child can be set correctly, and can be set as having
# been examined
self.family.set_child()
self.family.set_child_examined()
self.assertTrue(self.family.children[0].is_analysed())
self.assertIsNone(self.family.child)
# add another child, and check that when we set the child, we now pick
# up this child since the other one has previously been examined
self.family.add_child("child2", 'dad', 'mom', 'female', '2',
"/home/child2.vcf")
self.family.set_child()
self.assertEqual(self.family.child, self.family.children[1])
# make sure that set_child_examined() doesn't default to None if we
# have children left to analyse
self.family.add_child("child3", 'dad', 'mom', 'female', '2',
"/home/child3.vcf")
self.family.set_child()
self.family.set_child_examined()
self.assertIsNotNone(self.family.child)
# and set child = None once we have analysed all the children
self.family.set_child()
self.family.set_child_examined()
self.assertIsNone(self.family.child)
def test_load_families(self):
""" check that load_families works correctly
"""
# construct a temporary family that will have the same sample IDs etc
# as for the one loaded from the ped file.
family = Family("fam_ID")
family.add_child("proband", 'dad', 'mom', 'F', '2', "/path/to/proband_vcf.gz")
family.add_mother("mom", '0', '0', 'F', '1', "/path/to/mom_vcf.gz")
family.add_father("dad", '0', '0', 'M', '1', "/path/to/dad_vcf.gz")
# load the ped file, and check that the load_families function returns
# the expected Family object
self.assertEqual(load_families(self.path), [family])
# add an extra family, with multiple sibs
self.tempfile.write("fam_ID2 proband2 dad2 mom2 F 2 /path/to/proband2_vcf.gz\n")
self.tempfile.write("fam_ID2 dad2 0 0 M 1 /path/to/dad2_vcf.gz\n")
self.tempfile.write("fam_ID2 mom2 0 0 F 1 /path/to/mom2_vcf.gz\n")
self.tempfile.write("fam_ID2 sib dad2 mom2 F 2 /path/to/sib_vcf.gz\n")
self.tempfile.flush()
# construct a temporary family that will have the same sample IDs etc
# as for the one loaded from the ped file.
fam2 = Family("fam_ID2")
fam2.add_child("proband2", 'dad2', 'mom2', 'F', '2', "/path/to/proband2_vcf.gz")
fam2.add_child("sib", 'dad2', 'mom2', 'F', '2', "/path/to/sib_vcf.gz")
fam2.add_mother("mom2", '0', '0', 'F', '1', "/path/to/mom2_vcf.gz")
fam2.add_father("dad2", '0', '0', 'M', '1', "/path/to/dad2_vcf.gz")
# load the ped file, and check that the load_families function returns
# the expected Families objects
self.assertEqual(sorted(load_families(self.path)), sorted([family, fam2]))
class TestFamily(unittest.TestCase):
"""
"""
def setUp(self):
""" define a default Person object
"""
ID = "fam_ID"
self.family = Family(ID)
def test_add_father(self):
""" test that add_father() works correctly
"""
ID = "parent_ID"
path = "/home/parent.vcf"
affected = "1"
gender = "1"
# check that adding a male father doesn't raise an error
self.family.add_father(ID, path, affected, gender)
# check that adding a father for a second time is fine, but adding
# a different father raises an error
self.family.add_father(ID, path, affected, gender)
with self.assertRaises(ValueError):
self.family.add_father("different_ID", path, affected, gender)
# check that adding a female father raises an error
self.setUp()
gender = "2"
with self.assertRaises(ValueError):
self.family.add_father(ID, path, affected, gender)
def test_add_mother(self):
""" test that add_mother() works correctly
"""
ID = "parent_ID"
path = "/home/parent.vcf"
affected = "1"
gender = "2"
# check that adding a female mother doesn't raise an error
self.family.add_mother(ID, path, affected, gender)
# check that adding a mother for a second time is fine, but adding
# a different mother raises an error
self.family.add_mother(ID, path, affected, gender)
with self.assertRaises(ValueError):
self.family.add_mother("different_ID", path, affected, gender)
# check that adding a male mother raises an error
self.setUp()
gender = "1"
with self.assertRaises(ValueError):
self.family.add_mother(ID, path, affected, gender)
def test_add_child(self):
""" check that add_child() works correctly
"""
# check that we can add one child
self.family.add_child("child1", "/home/child1.vcf", "2", "1")
self.assertEqual(len(self.family.children), 1)
# check that adding multiple children works correctly
self.family.add_child("child2", "/home/child2.vcf", "2", "2")
self.family.add_child("child3", "/home/child3.vcf", "2", "1")
self.assertEqual(len(self.family.children), 3)
def test_set_child(self):
""" test that set_child() works correctly
"""
# add one child
self.family.add_child("child1", "/home/child1.vcf", "2", "1")
# check that the child can be set correctly
self.family.set_child()
self.assertEqual(self.family.child, self.family.children[0])
# add more children
self.family.add_child("child2", "/home/child2.vcf", "2", "1")
self.family.add_child("child3", "/home/child3.vcf", "2", "2")
# check that the child can be set correctly with multiple children
self.family.set_child()
self.assertIn(self.family.child, self.family.children)
def test_set_child_examined(self):
""" test that set_child_examined() works correctly
"""
# add one child
self.family.add_child("child1", "/home/child1.vcf", "2", "1")
# check that the child can be set correctly, and can be set as having
# been examined
self.family.set_child()
self.family.set_child_examined()
self.assertTrue(self.family.children[0].is_analysed())
self.assertIsNone(self.family.child)
# add another child, and check that when we set the child, we now pick
# up this child since the other one has previously been examined
self.family.add_child("child2", "/home/child2.vcf", "2", "2")
self.family.set_child()
self.assertEqual(self.family.child, self.family.children[1])
# make sure that set_child_examined() doesn't default to None if we
# have children left to analyse
self.family.add_child("child3", "/home/child3.vcf", "2", "2")
self.family.set_child()
self.family.set_child_examined()
self.assertIsNotNone(self.family.child)
# and set child = None once we have analysed all the children
self.family.set_child()
self.family.set_child_examined()
self.assertIsNone(self.family.child)
def test_analyse_trio(self):
''' test that analyse_trio() works correctly
'''
# construct the VCFs for the trio members
paths = {}
for member in ['child', 'mom', 'dad']:
vcf = make_vcf_header()
geno, pp_dnm = '0/0', ''
if member == 'child':
geno, pp_dnm = '0/1', ';DENOVO-SNP;PP_DNM=1'
vcf.append(
make_vcf_line(genotype=geno, extra='HGNC=ARID1B' + pp_dnm))
# write the VCF data to a file
handle = tempfile.NamedTemporaryFile(dir=self.temp_dir,
delete=False,
suffix='.vcf')
for x in vcf:
handle.write(x.encode('utf8'))
handle.flush()
paths[member] = handle.name
# create a Family object, so we can load the data from the trio's VCFs
fam_id = 'fam01'
child = Person(fam_id, 'child', 'dad', 'mom', 'female', '2',
paths['child'])
mom = Person(fam_id, 'mom', '0', '0', 'female', '1', paths['mom'])
dad = Person(fam_id, 'dad', '0', '0', 'male', '1', paths['dad'])
family = Family(fam_id, [child], mom, dad)
self.assertEqual(self.finder.analyse_trio(family), [(TrioGenotypes(
chrom="1",
pos=1,
child=SNV(
chrom="1",
position=1,
id=".",
ref="G",
alts="T",
qual='1000',
filter="PASS",
info="CQ=missense_variant;DENOVO-SNP;HGNC=ARID1B;PP_DNM=1",
format="DP:GT",
sample="50:0/1",
gender="female",
mnv_code=None),
mother=SNV(chrom="1",
position=1,
id=".",
ref="G",
alts="T",
qual='1000',
filter="PASS",
info="CQ=missense_variant;HGNC=ARID1B",
format="DP:GT",
sample="50:0/0",
gender="female",
mnv_code=None),
father=SNV(chrom="1",
position=1,
id=".",
ref="G",
alts="T",
qual='1000',
filter="PASS",
info="CQ=missense_variant;HGNC=ARID1B",
format="DP:GT",
sample="50:0/0",
gender="male",
mnv_code=None)), ['single_variant'], [
'Monoallelic', 'Mosaic'
], ['ARID1B'])])
def test_find_variants(self):
""" test that find_variants() works correctly
"""
# define the trio, so that we can know whether the parents are affected.
# The child also needs to be included and set, so that we can get the
# child ID for logging purposes.
family = Family("famID")
family.add_child("child_id", 'dad_id', 'mom_id', 'f', '2', "/vcf/path")
family.add_father("dad_id", '0', '0', 'm', '1', "/vcf/path")
family.add_mother("mom_id", '0', '0', 'f', '1', "/vcf/path")
family.set_child()
# create variants that cover various scenarios
snv1 = create_variant("F", "missense_variant|missense_variant",
"TEST1|TEST2")
snv2 = create_variant("F", "missense_variant|synonymous_variant",
"OTHER1|OTHER2")
snv3 = create_variant("F", "missense_variant", "")
snv4 = create_variant("F", "missense_variant", "TESTX", chrom="X")
self.finder.known_genes = {
"TEST1": {
"inh": ["Monoallelic"]
},
"OTHER1": {
"inh": ["Monoallelic"]
},
"OTHER2": {
"inh": ["Monoallelic"]
},
"TESTX": {
"inh": ["X-linked dominant"]
}
}
# check the simplest case, a variant in a known gene
self.assertEqual(
self.finder.find_variants([snv1], "TEST1", family),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST1"])])
# check that a gene not in a known gene does not pass
self.assertEqual(self.finder.find_variants([snv1], "TEST2", family),
[])
# check a variant where the gene is known, but the consequence for that
# gene is not functional, does not pass
self.assertEqual(self.finder.find_variants([snv2], "OTHER2", family),
[])
# check that intergenic variants (which lack HGNC symbols) do not pass
self.assertEqual(self.finder.find_variants([snv3], None, family), [])
# check that a variant on chrX passes through the allosomal instance
self.assertEqual(
self.finder.find_variants([snv4], "TESTX", family),
[(snv4, ["single_variant"], ["X-linked dominant"], ["TESTX"])])
# remove the known genes, so that the variants in unknown genes pass
self.finder.known_genes = None
self.assertEqual(
sorted(self.finder.find_variants([snv1], "TEST2", family)),
[(snv1, ["single_variant"], ["Monoallelic"], ["TEST2"]),
(snv1, ["single_variant"], ["Mosaic"], ["TEST2"])])
# but variants without gene symbols still are excluded
self.assertEqual(self.finder.find_variants([snv3], None, family), [])