def test_caching_cachedir():
vcf_fn = 'fixture/sample.vcf.gz'
cachedir = 'fixture/custom.vcfnp_cache/foo'
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn,
array_type='variants',
cachedir=cachedir)
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = variants(vcf_fn, cache=True, verbose=True, cachedir=cachedir)
a2 = np.load(cache_fn)
assert np.all(a == a2)
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn,
array_type='calldata',
cachedir=cachedir)
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = calldata(vcf_fn, cache=True, verbose=True, cachedir=cachedir)
a2 = np.load(cache_fn)
assert np.all(a == a2)
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn,
array_type='calldata_2d',
cachedir=cachedir)
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = calldata_2d(vcf_fn, cache=True, verbose=True, cachedir=cachedir)
a2 = np.load(cache_fn)
assert np.all(a == a2)
def test_condition():
v = variants('fixture/sample.vcf')
eq_(9, len(v))
c = calldata('fixture/sample.vcf', condition=v['FILTER']['PASS'])
eq_(5, len(c))
vf = variants('fixture/sample.vcf', condition=v['FILTER']['PASS'])
eq_(5, len(vf))
def test_caching_cachedir():
vcf_fn = 'fixture/sample.vcf.gz'
cachedir = 'fixture/custom.vcfnp_cache/foo'
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn, array_type='variants',
cachedir=cachedir)
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = variants(vcf_fn, cache=True, verbose=True, cachedir=cachedir)
a2 = np.load(cache_fn)
assert_array_equal(a, a2)
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn, array_type='calldata',
cachedir=cachedir)
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = calldata(vcf_fn, cache=True, verbose=True, cachedir=cachedir)
a2 = np.load(cache_fn)
assert_array_equal(a, a2)
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn, array_type='calldata_2d',
cachedir=cachedir)
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = calldata_2d(vcf_fn, cache=True, verbose=True, cachedir=cachedir)
a2 = np.load(cache_fn)
assert_array_equal(a, a2)
def test_condition():
v = variants('fixture/sample.vcf')
eq_(9, len(v))
c = calldata('fixture/sample.vcf', condition=v['FILTER']['PASS'])
eq_(5, len(c))
vf = variants('fixture/sample.vcf', condition=v['FILTER']['PASS'])
eq_(5, len(vf))
def test_caching_compression():
vcf_fn = 'fixture/sample.vcf.gz'
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn, array_type='variants',
compress=True)
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = variants(vcf_fn, cache=True, compress_cache=True, verbose=True)
a2 = np.load(cache_fn)['data']
assert_array_equal(a, a2)
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn, array_type='calldata',
compress=True)
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = calldata(vcf_fn, cache=True, compress_cache=True, verbose=True)
a2 = np.load(cache_fn)['data']
assert_array_equal(a, a2)
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn, array_type='calldata_2d',
compress=True)
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = calldata_2d(vcf_fn, cache=True, compress_cache=True, verbose=True)
a2 = np.load(cache_fn)['data']
assert_array_equal(a, a2)
def test_caching():
vcf_fn = 'fixture/sample.vcf.gz'
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn, array_type='variants')
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = variants(vcf_fn, cache=True, verbose=True)
a2 = np.load(cache_fn)
assert np.all(a == a2)
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn, array_type='calldata')
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = calldata(vcf_fn, cache=True, verbose=True)
a2 = np.load(cache_fn)
assert np.all(a == a2)
cache_fn = vcfnp.array._mk_cache_fn(vcf_fn, array_type='calldata_2d')
debug(cache_fn)
if os.path.exists(cache_fn):
os.remove(cache_fn)
a = calldata_2d(vcf_fn, cache=True, verbose=True)
a2 = np.load(cache_fn)
assert np.all(a == a2)
def test_variable_calldata():
c = calldata('fixture/test1.vcf')
eq_((1, 0), tuple(c['test2']['AD'][0]))
eq_((1, 0), tuple(c['test2']['AD'][1]))
eq_((1, 0), tuple(c['test2']['AD'][2]))
eq_(b'.', c['test2']['GT'][0])
eq_(b'0', c['test2']['GT'][1])
eq_(b'1', c['test2']['GT'][2])
def test_variable_calldata():
c = calldata('fixture/test1.vcf')
eq_((1, 0), tuple(c['test2']['AD'][0]))
eq_((1, 0), tuple(c['test2']['AD'][1]))
eq_((1, 0), tuple(c['test2']['AD'][2]))
eq_(b'.', c['test2']['GT'][0])
eq_(b'0', c['test2']['GT'][1])
eq_(b'1', c['test2']['GT'][2])
def test_calldata():
a = calldata('fixture/sample.vcf')
debug(repr(a))
eq_(b'0|0', a[0]['NA00001']['GT'])
eq_(True, a[0]['NA00001']['is_called'])
eq_(True, a[0]['NA00001']['is_phased'])
eq_((0, 0), tuple(a[0]['NA00001']['genotype']))
eq_((-1, -1), tuple(a[6]['NA00003']['genotype']))
eq_((-1, -1), tuple(a[7]['NA00003']['genotype']))
eq_((10, 10), tuple(a[0]['NA00001']['HQ']))
def test_calldata():
a = calldata('fixture/sample.vcf')
debug(repr(a))
eq_(b'0|0', a[0]['NA00001']['GT'])
eq_(True, a[0]['NA00001']['is_called'])
eq_(True, a[0]['NA00001']['is_phased'])
eq_((0, 0), tuple(a[0]['NA00001']['genotype']))
eq_((-1, -1), tuple(a[6]['NA00003']['genotype']))
eq_((-1, -1), tuple(a[7]['NA00003']['genotype']))
eq_((10, 10), tuple(a[0]['NA00001']['HQ']))
def test_truncate():
# https://github.com/alimanfoo/vcfnp/issues/54
vcf_fn = 'fixture/test54.vcf.gz'
# truncate by default
v = variants(vcf_fn, region='chr1:10-100')
eq_(2, len(v))
c = calldata(vcf_fn, region='chr1:10-100')
eq_(2, len(c))
c2d = calldata_2d(vcf_fn, region='chr1:10-100')
eq_(2, len(c2d))
# don't truncate
v = variants(vcf_fn, region='chr1:10-100', truncate=False)
eq_(3, len(v))
c = calldata(vcf_fn, region='chr1:10-100', truncate=False)
eq_(3, len(c))
c2d = calldata_2d(vcf_fn, region='chr1:10-100', truncate=False)
eq_(3, len(c2d))
def test_truncate():
# https://github.com/alimanfoo/vcfnp/issues/54
vcf_fn = 'fixture/test54.vcf.gz'
# truncate by default
v = variants(vcf_fn, region='chr1:10-100')
eq_(2, len(v))
c = calldata(vcf_fn, region='chr1:10-100')
eq_(2, len(c))
c2d = calldata_2d(vcf_fn, region='chr1:10-100')
eq_(2, len(c2d))
# don't truncate
v = variants(vcf_fn, region='chr1:10-100', truncate=False)
eq_(3, len(v))
c = calldata(vcf_fn, region='chr1:10-100', truncate=False)
eq_(3, len(c))
c2d = calldata_2d(vcf_fn, region='chr1:10-100', truncate=False)
eq_(3, len(c2d))
def test_missing_calldata_cleared():
c = calldata('fixture/test32.vcf')['AC0093-C']
# first variant, non-missing
eq_(b'0/0', c['GT'][0])
eq_((0, 0), tuple(c['genotype'][0]))
eq_(8, c['DP'][0])
eq_(3, c['GQ'][0])
# second variant, missing
eq_(b'./.', c['GT'][1])
eq_((-1, -1), tuple(c['genotype'][1]))
eq_(0, c['DP'][1]) # should be default fill value
eq_(0, c['GQ'][1]) # should be default fill value
def test_missing_calldata_cleared():
c = calldata('fixture/test32.vcf')['AC0093-C']
# first variant, non-missing
eq_(b'0/0', c['GT'][0])
eq_((0, 0), tuple(c['genotype'][0]))
eq_(8, c['DP'][0])
eq_(3, c['GQ'][0])
# second variant, missing
eq_(b'./.', c['GT'][1])
eq_((-1, -1), tuple(c['genotype'][1]))
eq_(0, c['DP'][1]) # should be default fill value
eq_(0, c['GQ'][1]) # should be default fill value
def test_missing_calldata():
c = calldata('fixture/test1.vcf')
# first variant, second sample
eq_(b'.', c['test2']['GT'][0])
eq_((-1, -1), tuple(c['test2']['genotype'][0]))
eq_((1, 0), tuple(c['test2']['AD'][0])) # data are present
# third variant, third sample
eq_(b'.', c['test3']['GT'][2])
eq_((-1, -1), tuple(c['test3']['genotype'][2]))
eq_((0, 0), tuple(c['test3']['AD'][2])) # default fill
# third variant, fourth sample
eq_(b'./.', c['test4']['GT'][2])
eq_((-1, -1), tuple(c['test4']['genotype'][2]))
eq_((0, 0), tuple(c['test4']['AD'][2])) # default fill
def test_missing_calldata():
c = calldata('fixture/test1.vcf')
# first variant, second sample
eq_(b'.', c['test2']['GT'][0])
eq_((-1, -1), tuple(c['test2']['genotype'][0]))
eq_((1, 0), tuple(c['test2']['AD'][0])) # data are present
# third variant, third sample
eq_(b'.', c['test3']['GT'][2])
eq_((-1, -1), tuple(c['test3']['genotype'][2]))
eq_((0, 0), tuple(c['test3']['AD'][2])) # default fill
# third variant, fourth sample
eq_(b'./.', c['test4']['GT'][2])
eq_((-1, -1), tuple(c['test4']['genotype'][2]))
eq_((0, 0), tuple(c['test4']['AD'][2])) # default fill
def test_calldata_region():
a = calldata('fixture/sample.vcf.gz', region='20')
eq_(6, len(a))
def test_missing_format_definition():
# FORMAT field DP not declared in VCF header
c = calldata('fixture/test14.vcf', fields=['DP'],
vcf_types={'DP': 'Integer'})
eq_(1, c[2]['NA00001']['DP'])
def test_duplicate_field_definitions():
variants('fixture/test10.vcf')
# should not raise, but print useful message to stderr
calldata('fixture/test10.vcf')
def test_duplicate_field_definitions():
variants('fixture/test10.vcf')
# should not raise, but print useful message to stderr
calldata('fixture/test10.vcf')
def test_missing_format_definition():
# FORMAT field DP not declared in VCF header
c = calldata('fixture/test14.vcf',
fields=['DP'],
vcf_types={'DP': 'Integer'})
eq_(1, c[2]['NA00001']['DP'])
def test_calldata_region_empty():
a = calldata('fixture/sample.vcf.gz', region='18')
eq_(0, len(a))
a = calldata('fixture/sample.vcf.gz', region='19:113-200')
eq_(0, len(a))
def test_calldata_region_empty():
a = calldata('fixture/sample.vcf.gz', region='18')
eq_(0, len(a))
a = calldata('fixture/sample.vcf.gz', region='19:113-200')
eq_(0, len(a))
def test_calldata_region():
a = calldata('fixture/sample.vcf.gz', region='20')
eq_(6, len(a))