def test_zoomify():
kwargs = dict(chunksize=10, nproc=1, columns=None, dtypes=None, agg=None)
with isolated_filesystem():
zoomify_cooler(op.join(datadir, "toy.asymm.2.cool"),
"test.2.mcool",
resolutions=[4, 8, 16, 32],
**kwargs)
for res in [2, 4, 8, 16, 32]:
cooler_cmp(
"test.2.mcool::resolutions/{}".format(res),
op.join(datadir, "toy.asymm.{}.cool".format(res)),
)
# include base resolution
zoomify_cooler(op.join(datadir, "toy.asymm.2.cool"),
"test.2.mcool",
resolutions=[2, 4, 8, 16, 32],
**kwargs)
for res in [2, 4, 8, 16, 32]:
cooler_cmp(
"test.2.mcool::resolutions/{}".format(res),
op.join(datadir, "toy.asymm.{}.cool".format(res)),
)
# impossible resolution to obtain
with pytest.raises(ValueError):
zoomify_cooler(op.join(datadir, "toy.asymm.2.cool"),
"test.2.mcool",
resolutions=[4, 5, 32],
**kwargs)
def test_zoomify():
kwargs = dict(chunksize=10, nproc=1, columns=None, dtypes=None, agg=None)
with isolated_filesystem():
zoomify_cooler(op.join(datadir, "toy.asymm.2.cool"),
"test.2.mcool",
resolutions=[4, 8, 16, 32],
**kwargs)
for res in [2, 4, 8, 16, 32]:
cooler_cmp(
"test.2.mcool::resolutions/{}".format(res),
op.join(datadir, "toy.asymm.{}.cool".format(res)),
)
def test_mv():
with isolated_filesystem():
ref_file = "test.ref.mcool"
src_file = "test.src.mcool"
shutil.copyfile(op.join(testdir, "data", "toy.symm.upper.2.mcool"),
ref_file)
shutil.copyfile(op.join(testdir, "data", "toy.symm.upper.2.mcool"),
src_file)
fileops.mv(src_file + "::resolutions/2", src_file + "::abc/d")
with h5py.File(src_file) as f:
assert "resolutions/2" not in f
assert "abc/d" in f
cooler_cmp(ref_file + "::resolutions/2", src_file + "::abc/d")
def test_mv():
with isolated_filesystem():
ref_file = 'test.ref.mcool'
src_file = 'test.src.mcool'
shutil.copyfile(op.join(testdir, 'data', 'toy.symm.upper.2.mcool'),
ref_file)
shutil.copyfile(op.join(testdir, 'data', 'toy.symm.upper.2.mcool'),
src_file)
fileops.mv(src_file + '::resolutions/2', src_file + '::abc/d')
with h5py.File(src_file) as f:
assert 'resolutions/2' not in f
assert 'abc/d' in f
cooler_cmp(ref_file + '::resolutions/2', src_file + '::abc/d')
def test_coarsen(input_uri, factor, ref_uri):
kwargs = dict(
chunksize=10,
nproc=1,
columns=None,
dtypes=None,
agg=None
)
with isolated_filesystem():
coarsen_cooler(
input_uri,
'test.cool',
factor,
**kwargs
)
cooler_cmp('test.cool', ref_uri)
def test_zoomify():
kwargs = dict(
chunksize=10,
nproc=1,
columns=None,
dtypes=None,
agg=None,
)
with isolated_filesystem():
zoomify_cooler(op.join(testdir, 'data', 'toy.asymm.2.cool'),
'test.2.mcool',
resolutions=[4, 8, 16, 32],
**kwargs)
for res in [2, 4, 8, 16, 32]:
cooler_cmp(
'test.2.mcool::resolutions/{}'.format(res),
op.join(testdir, 'data', 'toy.asymm.{}.cool'.format(res)))
def test_cp():
with isolated_filesystem():
src_file = op.join(testdir, "data", "toy.symm.upper.2.mcool")
# file-to-file
src_uri = src_file + "::resolutions/2"
fileops.cp(src_uri, "test.2.cool")
cooler_cmp(src_uri, "test.2.cool")
# within-file
test_file = "test.src.mcool"
shutil.copyfile(src_file, test_file)
fileops.cp(test_file + "::resolutions/2", test_file + "::abc/d")
cooler_cmp(test_file + "::resolutions/2", test_file + "::abc/d")
with h5py.File(test_file) as f:
assert "resolutions/2" in f
assert "abc/d" in f
assert f["resolutions/2"].id != f["abc/d"].id
def test_cp():
with isolated_filesystem() as fs:
src_file = op.join(testdir, 'data', 'toy.symm.upper.2.mcool')
# file-to-file
src_uri = src_file + '::resolutions/2'
fileops.cp(src_uri, 'test.2.cool')
cooler_cmp(src_uri, 'test.2.cool')
# within-file
test_file = 'test.src.mcool'
shutil.copyfile(src_file, test_file)
fileops.cp(test_file + '::resolutions/2', test_file + '::abc/d')
cooler_cmp(test_file + '::resolutions/2', test_file + '::abc/d')
with h5py.File(test_file) as f:
assert 'resolutions/2' in f
assert 'abc/d' in f
assert f['resolutions/2'].id != f['abc/d'].id
def test_coarsen(input_uri, factor, ref_uri):
with isolated_filesystem():
kwargs = dict(chunksize=10,
nproc=1,
columns=None,
dtypes=None,
agg=None)
coarsen_cooler(input_uri, "test.cool", factor, **kwargs)
cooler_cmp("test.cool", ref_uri)
# custom dtype
kwargs = dict(chunksize=10,
nproc=1,
columns=None,
dtypes={'count': np.float64})
coarsen_cooler(input_uri, "test.cool", factor, **kwargs)
with h5py.File('test.cool', 'r') as f:
assert f['pixels/count'].dtype.kind == 'f'
# custom aggregator
kwargs = dict(chunksize=10,
nproc=1,
columns=None,
dtypes=None,
agg={'count': 'mean'})
coarsen_cooler(input_uri, "test.cool", factor, **kwargs)
# parallel
kwargs = dict(chunksize=10,
nproc=2,
columns=None,
dtypes=None,
agg=None)
coarsen_cooler(input_uri, "test.cool", factor, **kwargs)
# raise on missing value column
kwargs = dict(chunksize=10,
nproc=2,
columns=['missing'],
dtypes=None,
agg=None)
with pytest.raises(ValueError):
coarsen_cooler(input_uri, "test.cool", factor, **kwargs)
def test_ln():
with isolated_filesystem():
src_file = op.join(testdir, "data", "toy.symm.upper.2.mcool")
# within-file hard link
test_file = "test.hardlink.mcool"
shutil.copyfile(src_file, test_file)
fileops.ln(test_file + "::resolutions/2", test_file + "::abc/d")
with h5py.File(test_file) as f:
assert "resolutions/2" in f
assert "abc/d" in f
assert f["resolutions/2"].id == f["abc/d"].id
cooler_cmp(test_file + "::resolutions/2", test_file + "::abc/d")
# within-file soft link
test_file = "test.softlink.mcool"
shutil.copyfile(src_file, test_file)
fileops.ln(test_file + "::resolutions/2",
test_file + "::abc/d",
soft=True)
with h5py.File(test_file) as f:
assert "resolutions/2" in f
assert "abc/d" in f
assert f["resolutions/2"].id == f["abc/d"].id
cooler_cmp(test_file + "::resolutions/2", test_file + "::abc/d")
# between-file external link
test_file = "test.extlink.mcool"
dst_file = "test.dst.cool"
shutil.copyfile(src_file, test_file)
fileops.ln(test_file + "::resolutions/2",
dst_file + "::abc/d",
soft=True)
cooler_cmp(test_file + "::resolutions/2", dst_file + "::abc/d")
def test_ln():
with isolated_filesystem() as fs:
src_file = op.join(testdir, 'data', 'toy.symm.upper.2.mcool')
# within-file hard link
test_file = 'test.hardlink.mcool'
shutil.copyfile(src_file, test_file)
fileops.ln(test_file + '::resolutions/2', test_file + '::abc/d')
with h5py.File(test_file) as f:
assert 'resolutions/2' in f
assert 'abc/d' in f
assert f['resolutions/2'].id == f['abc/d'].id
cooler_cmp(test_file + '::resolutions/2', test_file + '::abc/d')
# within-file soft link
test_file = 'test.softlink.mcool'
shutil.copyfile(src_file, test_file)
fileops.ln(test_file + '::resolutions/2',
test_file + '::abc/d',
soft=True)
with h5py.File(test_file) as f:
assert 'resolutions/2' in f
assert 'abc/d' in f
assert f['resolutions/2'].id == f['abc/d'].id
cooler_cmp(test_file + '::resolutions/2', test_file + '::abc/d')
# between-file external link
test_file = 'test.extlink.mcool'
dst_file = 'test.dst.cool'
shutil.copyfile(src_file, test_file)
fileops.ln(test_file + '::resolutions/2',
dst_file + '::abc/d',
soft=True)
cooler_cmp(test_file + '::resolutions/2', dst_file + '::abc/d')
def test_dump():
runner = CliRunner()
with runner.isolated_filesystem():
f_in = op.join(datadir, "toy.symm.upper.2.cool")
result = runner.invoke(dump, [f_in])
assert result.exit_code == 0
# roundtrip symm-upper data
bins = pd.read_csv(
StringIO(runner.invoke(dump, [f_in, "-H", "-t", "bins"]).output), sep="\t"
)
pixels = pd.read_csv(
StringIO(runner.invoke(dump, [f_in, "-H"]).output), sep="\t"
)
cooler.create_cooler("out.cool", bins, pixels, symmetric_upper=True)
cooler_cmp(f_in, "out.cool")
# duplexed output
pixels2 = pd.read_csv(
StringIO(runner.invoke(dump, [f_in, "--matrix", "-H"]).output), sep="\t"
)
assert len(pixels2) > len(pixels)
upper = pixels2[pixels2["bin1_id"] <= pixels2["bin2_id"]].reset_index(drop=True)
assert np.allclose(pixels, upper)
# lower triangle
trans_lower = pd.read_csv(
StringIO(
runner.invoke(dump, [f_in, "-H", "-r", "chr2", "-r2", "chr1"]).output
),
sep="\t",
)
assert len(trans_lower) == 0
trans_lower = pd.read_csv(
StringIO(
runner.invoke(
dump, [f_in, "-m", "-H", "-r", "chr2", "-r2", "chr1"]
).output
),
sep="\t",
)
assert len(trans_lower) > 0
# roundtrip square data
f_in = op.join(datadir, "toy.asymm.2.cool")
bins = pd.read_csv(
StringIO(runner.invoke(dump, [f_in, "-H", "-t", "bins"]).output), sep="\t"
)
pixels = pd.read_csv(
StringIO(runner.invoke(dump, [f_in, "-H"]).output), sep="\t"
)
cooler.create_cooler("out.cool", bins, pixels, symmetric_upper=False)
cooler_cmp(f_in, "out.cool")
pixels2 = pd.read_csv(
StringIO(runner.invoke(dump, [f_in, "--matrix", "-H"]).output), sep="\t"
)
assert np.allclose(pixels, pixels2)
# for square data, -m is a no-op
lower1 = pd.read_csv(
StringIO(
runner.invoke(dump, [f_in, "-H", "-r", "chr2", "-r2", "chr1"]).output
),
sep="\t",
)
lower2 = pd.read_csv(
StringIO(
runner.invoke(
dump, [f_in, "-m", "-H", "-r", "chr2", "-r2", "chr1"]
).output
),
sep="\t",
)
assert np.allclose(lower1, lower2)
def test_dump():
runner = CliRunner()
with runner.isolated_filesystem():
f_in = op.join(datadir, 'toy.symm.upper.2.cool')
result = runner.invoke(dump, [
f_in,
])
assert result.exit_code == 0
# roundtrip symm-upper data
bins = pd.read_csv(StringIO(
runner.invoke(dump, [f_in, '-H', '-t', 'bins']).output),
sep='\t')
pixels = pd.read_csv(StringIO(
runner.invoke(dump, [f_in, '-H']).output),
sep='\t')
cooler.create_cooler('out.cool', bins, pixels, symmetric_upper=True)
cooler_cmp(f_in, 'out.cool')
# duplexed output
pixels2 = pd.read_csv(StringIO(
runner.invoke(dump, [f_in, '--matrix', '-H']).output),
sep='\t')
assert len(pixels2) > len(pixels)
upper = pixels2[pixels2['bin1_id'] <= pixels2['bin2_id']].reset_index(
drop=True)
assert np.allclose(pixels, upper)
# lower triangle
trans_lower = pd.read_csv(StringIO(
runner.invoke(dump,
[f_in, '-H', '-r', 'chr2', '-r2', 'chr1']).output),
sep='\t')
assert len(trans_lower) == 0
trans_lower = pd.read_csv(StringIO(
runner.invoke(
dump, [f_in, '-m', '-H', '-r', 'chr2', '-r2', 'chr1']).output),
sep='\t')
assert len(trans_lower) > 0
# roundtrip square data
f_in = op.join(datadir, 'toy.asymm.2.cool')
bins = pd.read_csv(StringIO(
runner.invoke(dump, [f_in, '-H', '-t', 'bins']).output),
sep='\t')
pixels = pd.read_csv(StringIO(
runner.invoke(dump, [f_in, '-H']).output),
sep='\t')
cooler.create_cooler('out.cool', bins, pixels, symmetric_upper=False)
cooler_cmp(f_in, 'out.cool')
pixels2 = pd.read_csv(StringIO(
runner.invoke(dump, [f_in, '--matrix', '-H']).output),
sep='\t')
assert np.allclose(pixels, pixels2)
# for square data, -m is a no-op
lower1 = pd.read_csv(StringIO(
runner.invoke(dump,
[f_in, '-H', '-r', 'chr2', '-r2', 'chr1']).output),
sep='\t')
lower2 = pd.read_csv(StringIO(
runner.invoke(
dump, [f_in, '-m', '-H', '-r', 'chr2', '-r2', 'chr1']).output),
sep='\t')
assert np.allclose(lower1, lower2)
def test_dump():
runner = CliRunner()
with runner.isolated_filesystem():
f_in = op.join(datadir, "toy.symm.upper.2.cool")
result = runner.invoke(dump, [f_in])
assert result.exit_code == 0
result = runner.invoke(dump, [f_in, "-t", "chroms", "--columns", "length"])
assert result.exit_code == 0
result = runner.invoke(dump, [f_in, "-t", "bins", "--columns", "chrom,start"])
assert result.exit_code == 0
result = runner.invoke(dump, [f_in, "-r", "chr1"])
assert result.exit_code == 0
result = runner.invoke(dump, [f_in, "-r", "chr1:0-16", "-r2", "chr1:10-25"])
assert result.exit_code == 0
result = runner.invoke(dump, [f_in, "-r", "chr1:10-25", "-r2", "chr1:0-5"])
assert result.exit_code == 0
result = runner.invoke(dump, [f_in, "--join"])
assert result.exit_code == 0
result = runner.invoke(dump, [f_in, "--join", "--one-based-ids"])
assert result.exit_code == 0
result = runner.invoke(dump, [f_in, "--join", "--one-based-starts"])
assert result.exit_code == 0
result = runner.invoke(dump, [f_in, "--annotate", "chrom", "--one-based-starts"])
assert result.exit_code == 0
# unbalanced file
result = runner.invoke(dump, [f_in, "-b"])
assert result.exit_code == 1
# roundtrip symm-upper data
result = runner.invoke(dump, [f_in, "-H", "-t", "bins"])
bins = pd.read_csv(StringIO(result.output), sep="\t")
result = runner.invoke(dump, [f_in, "-H"])
pixels = pd.read_csv(StringIO(result.output), sep="\t")
cooler.create_cooler("out.cool", bins, pixels, symmetric_upper=True)
cooler_cmp(f_in, "out.cool")
# duplexed output
result = runner.invoke(dump, [f_in, "--matrix", "-H"])
pixels2 = pd.read_csv(StringIO(result.output), sep="\t")
assert len(pixels2) > len(pixels)
upper = pixels2[pixels2["bin1_id"] <= pixels2["bin2_id"]].reset_index(drop=True)
assert np.allclose(pixels, upper)
# lower triangle
result = runner.invoke(dump, [f_in, "-H", "-r", "chr2", "-r2", "chr1"])
trans_lower = pd.read_csv(StringIO(result.output), sep="\t")
assert len(trans_lower) == 0
result = runner.invoke(dump, [f_in, "-m", "-H", "-r", "chr2", "-r2", "chr1"])
trans_lower = pd.read_csv(StringIO(result.output), sep="\t")
assert len(trans_lower) > 0
# roundtrip square data
f_in = op.join(datadir, "toy.asymm.2.cool")
result = runner.invoke(dump, [f_in, "-H", "-t", "bins"])
bins = pd.read_csv(StringIO(result.output), sep="\t")
result = runner.invoke(dump, [f_in, "-H"])
pixels = pd.read_csv(StringIO(result.output), sep="\t")
cooler.create_cooler("out.cool", bins, pixels, symmetric_upper=False)
cooler_cmp(f_in, "out.cool")
result = runner.invoke(dump, [f_in, "--matrix", "-H"])
pixels2 = pd.read_csv(StringIO(result.output), sep="\t")
assert np.allclose(pixels, pixels2)
# for square data, -m is a no-op
result = runner.invoke(dump, [f_in, "-H", "-r", "chr2", "-r2", "chr1"])
lower1 = pd.read_csv(StringIO(result.output), sep="\t")
result = runner.invoke(dump, [f_in, "-m", "-H", "-r", "chr2", "-r2", "chr1"])
lower2 = pd.read_csv(StringIO(result.output), sep="\t")
assert np.allclose(lower1, lower2)
