def test_count_value_per_row(self):
mat = numpy.array([[0, 0], [1, -1], [2, -1], [-1, -1]])
missing_counter = row_value_counter_fact(value=-1)
assert numpy.all(missing_counter(mat) == [0, 1, 1, 2])
missing_counter = row_value_counter_fact(value=-1, ratio=True)
assert numpy.allclose(missing_counter(mat), [0., 0.5, 0.5, 1.])
hdf5 = VariationsH5(join(TEST_DATA_DIR, '1000snps.hdf5'), mode='r')
chunks = list(hdf5.iterate_chunks())
gt_chunk = first(select_dset_from_chunks(chunks, '/calls/GT'))
homo_counter = row_value_counter_fact(value=2)
assert numpy.all(homo_counter(gt_chunk) == [0, 0, 4, 0, 1])
missing_counter = row_value_counter_fact(value=2, ratio=True)
expected = [0., 0, 0.66666, 0., 0.166666]
assert numpy.allclose(missing_counter(gt_chunk), expected)
hdf5.close()
def test_count_value_per_row(self):
mat = numpy.array([[0, 0], [1, -1], [2, -1], [-1, -1]])
missing_counter = row_value_counter_fact(value=-1)
assert numpy.all(missing_counter(mat) == [0, 1, 1, 2])
missing_counter = row_value_counter_fact(value=-1, ratio=True)
assert numpy.allclose(missing_counter(mat), [0., 0.5, 0.5, 1.])
hdf5 = VariationsH5(join(TEST_DATA_DIR, '1000snps.hdf5'), mode='r')
chunks = list(hdf5.iterate_chunks())
gt_chunk = first(select_dset_from_chunks(chunks, '/calls/GT'))
homo_counter = row_value_counter_fact(value=2)
assert numpy.all(homo_counter(gt_chunk) == [0, 0, 4, 0, 1])
missing_counter = row_value_counter_fact(value=2, ratio=True)
expected = [0., 0, 0.66666, 0., 0.166666]
assert numpy.allclose(missing_counter(gt_chunk), expected)
hdf5.close()
def _count_gts(variations):
if variations[GT_FIELD].shape[0] == 0:
return numpy.array([]), numpy.array([]), numpy.array([])
gts = variations[GT_FIELD]
gts = gts[...]
# get rid of genotypes with missing alleles
missing_alleles = gts == MISSING_INT
miss_gts = numpy.any(missing_alleles, axis=2)
# We pack the genotype of a sample that is in third axes as two
# integers as one integer: 1, 1 -> 11 0, 1 -> 01, 0, 0-> 0
gts_per_haplo = [(gts[:, :, idx].astype(numpy.int16)) * (100**idx)
for idx in range(gts.shape[2])]
packed_gts = None
for gts_ in gts_per_haplo:
if packed_gts is None:
packed_gts = gts_
else:
packed_gts += gts_
packed_gts[miss_gts] = MISSING_INT
different_gts = numpy.unique(packed_gts)
# Count genotypes, h**o, het and alleles
ploidy = gts.shape[2]
counts = {}
for gt in different_gts:
count_gt_by_row = row_value_counter_fact(gt)
gt_counts = count_gt_by_row(packed_gts)
if gt == MISSING_INT:
continue
unpacked_gt = _packed_gt_to_tuple(gt, ploidy)
if unpacked_gt not in counts:
counts[unpacked_gt] = gt_counts
else:
counts[unpacked_gt] += gt_counts
return counts
def _count_gts(variations):
if variations[GT_FIELD].shape[0] == 0:
return numpy.array([]), numpy.array([]), numpy.array([])
gts = variations[GT_FIELD]
gts = gts[...]
# get rid of genotypes with missing alleles
missing_alleles = gts == MISSING_INT
miss_gts = numpy.any(missing_alleles, axis=2)
# We pack the genotype of a sample that is in third axes as two
# integers as one integer: 1, 1 -> 11 0, 1 -> 01, 0, 0-> 0
gts_per_haplo = [(gts[:, :, idx].astype(numpy.int16)) * (100 ** idx) for idx in range(gts.shape[2])]
packed_gts = None
for gts_ in gts_per_haplo:
if packed_gts is None:
packed_gts = gts_
else:
packed_gts += gts_
packed_gts[miss_gts] = MISSING_INT
different_gts = numpy.unique(packed_gts)
# Count genotypes, h**o, het and alleles
ploidy = gts.shape[2]
counts = {}
for gt in different_gts:
count_gt_by_row = row_value_counter_fact(gt)
gt_counts = count_gt_by_row(packed_gts)
if gt == MISSING_INT:
continue
unpacked_gt = _packed_gt_to_tuple(gt, ploidy)
if unpacked_gt not in counts:
counts[unpacked_gt] = gt_counts
else:
counts[unpacked_gt] += gt_counts
return counts
