def test_load_zero_one_based_bg2():
kwargs = dict(
format="bg2",
metadata=None,
assembly="toy",
chunksize=10,
field=(),
count_as_float=False,
comment_char="#",
input_copy_status="unique",
no_symmetric_upper=False,
storage_options=None,
)
# 1-based-start BG2 input
ref = "toy.symm.upper.1.ob.bg2"
bins_path = op.join(testdir, "data", "toy.chrom.sizes") + ":1"
pixels_path = op.join(testdir, "data", ref)
load.callback(bins_path,
pixels_path,
testcool_path,
one_based=True,
**kwargs)
# reference, 1-based starts
ref_df = pd.read_csv(
pixels_path,
sep="\t",
names=[
"chrom1", "start1", "end1", "chrom2", "start2", "end2", "count"
],
)
# output
out_df = cooler.Cooler(testcool_path).pixels(join=True)[:]
out_df["start1"] += 1
out_df["start2"] += 1
assert np.all(out_df == ref_df)
# 0-based-start BG2 input
ref = "toy.symm.upper.1.zb.bg2"
bins_path = op.join(testdir, "data", "toy.chrom.sizes") + ":1"
pixels_path = op.join(testdir, "data", ref)
load.callback(bins_path,
pixels_path,
testcool_path,
one_based=False,
**kwargs)
# reference, 0-based starts
ref_df = pd.read_csv(
pixels_path,
sep="\t",
names=[
"chrom1", "start1", "end1", "chrom2", "start2", "end2", "count"
],
)
# output
out_df = cooler.Cooler(testcool_path).pixels(join=True)[:]
assert np.all(out_df == ref_df)
def test_load_zero_one_based_bg2():
kwargs = dict(format='bg2',
metadata=None,
assembly='toy',
chunksize=10,
field=(),
count_as_float=False,
comment_char='#',
input_copy_status='unique',
no_symmetric_upper=False,
storage_options=None)
# 1-based-start BG2 input
ref = 'toy.symm.upper.1.ob.bg2'
bins_path = op.join(testdir, 'data', 'toy.chrom.sizes') + ':1'
pixels_path = op.join(testdir, 'data', ref)
load.callback(bins_path,
pixels_path,
testcool_path,
one_based=True,
**kwargs)
# reference, 1-based starts
ref_df = pd.read_csv(pixels_path,
sep='\t',
names=[
'chrom1', 'start1', 'end1', 'chrom2', 'start2',
'end2', 'count'
])
# output
out_df = cooler.Cooler(testcool_path).pixels(join=True)[:]
out_df['start1'] += 1
out_df['start2'] += 1
assert np.all(out_df == ref_df)
# 0-based-start BG2 input
ref = 'toy.symm.upper.1.zb.bg2'
bins_path = op.join(testdir, 'data', 'toy.chrom.sizes') + ':1'
pixels_path = op.join(testdir, 'data', ref)
load.callback(bins_path,
pixels_path,
testcool_path,
one_based=False,
**kwargs)
# reference, 0-based starts
ref_df = pd.read_csv(pixels_path,
sep='\t',
names=[
'chrom1', 'start1', 'end1', 'chrom2', 'start2',
'end2', 'count'
])
# output
out_df = cooler.Cooler(testcool_path).pixels(join=True)[:]
assert np.all(out_df == ref_df)
def test_load_zero_one_based_coo():
kwargs = dict(
format='coo',
metadata=None,
assembly='toy',
chunksize=10,
field=(),
count_as_float=False,
comment_char='#',
symmetric_input='unique',
no_symmetric_storage=False,
storage_options=None
)
# 1-based-start COO input
ref = 'toy.symm.upper.1.ob.coo'
bins_path = op.join(testdir, 'data', 'toy.chrom.sizes') + ':1'
pixels_path = op.join(testdir, 'data', ref)
load.callback(
bins_path,
pixels_path,
testcool_path,
one_based=True,
**kwargs
)
# reference, 1-based starts
ref_df = pd.read_table(
pixels_path,
names=['bin1_id', 'bin2_id', 'count'])
# output
out_df = cooler.Cooler(testcool_path).pixels()[:]
out_df['bin1_id'] += 1
out_df['bin2_id'] += 1
assert np.all(out_df == ref_df)
# 0-based-start COO input
ref = 'toy.symm.upper.1.zb.coo'
bins_path = op.join(testdir, 'data', 'toy.chrom.sizes') + ':1'
pixels_path = op.join(testdir, 'data', ref)
load.callback(
bins_path,
pixels_path,
testcool_path,
one_based=False,
**kwargs
)
# reference, 0-based starts
ref_df = pd.read_table(
pixels_path,
names=['bin1_id', 'bin2_id', 'count'])
# output
out_df = cooler.Cooler(testcool_path).pixels()[:]
assert np.all(out_df == ref_df)
def test_load_bg2_vs_coo():
kwargs = dict(
metadata=None,
assembly='hg19',
chunksize=int(20e6),
field=(),
count_as_float=False,
one_based=False,
comment_char='#',
symmetric_input='unique',
no_symmetric_storage=False,
storage_options=None,
)
out_path1 = op.join(tmp, 'test1.cool')
out_path2 = op.join(tmp, 'test2.cool')
load.callback(
op.join(testdir, 'data', 'hg19.bins.2000kb.bed.gz'),
op.join(testdir, 'data', 'hg19.GM12878-MboI.matrix.2000kb.bg2.gz'),
out_path1,
format='bg2',
**kwargs
)
load.callback(
op.join(testdir, 'data', 'hg19.bins.2000kb.bed.gz'),
op.join(testdir, 'data', 'hg19.GM12878-MboI.matrix.2000kb.coo.txt'),
out_path2,
format='coo',
**kwargs
)
with h5py.File(out_path1, 'r') as f1, \
h5py.File(out_path2, 'r') as f2:
for col in ['bin1_id', 'bin2_id', 'count']:
assert np.all(f1['pixels'][col][:] == f2['pixels'][col][:])
for fp in [out_path1, out_path2]:
try:
os.remove(fp)
except OSError:
pass
def test_load_bg2_vs_coo():
kwargs = dict(
metadata=None,
assembly="hg19",
chunksize=int(20e6),
field=(),
count_as_float=False,
one_based=False,
comment_char="#",
input_copy_status="unique",
no_symmetric_upper=False,
storage_options=None,
)
out_path1 = op.join(tmp, "test1.cool")
out_path2 = op.join(tmp, "test2.cool")
load.callback(op.join(testdir, "data", "hg19.bins.2000kb.bed.gz"),
op.join(testdir, "data",
"hg19.GM12878-MboI.matrix.2000kb.bg2.gz"),
out_path1,
format="bg2",
**kwargs)
load.callback(op.join(testdir, "data", "hg19.bins.2000kb.bed.gz"),
op.join(testdir, "data",
"hg19.GM12878-MboI.matrix.2000kb.coo.txt"),
out_path2,
format="coo",
**kwargs)
with h5py.File(out_path1, "r") as f1, h5py.File(out_path2, "r") as f2:
for col in ["bin1_id", "bin2_id", "count"]:
assert np.all(f1["pixels"][col][:] == f2["pixels"][col][:])
for fp in [out_path1, out_path2]:
try:
os.remove(fp)
except OSError:
pass