def test_list_input(self):
"""
Check AJIVE can take a list input.
"""
ajive = AJIVE(init_signal_ranks=[2, 3])
ajive.fit(Xs=[self.X, self.Y])
self.assertTrue(set(ajive.block_names) == set([0, 1]))
def test_indiv(data):
dat = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.fit(Xs=dat)
blocks = ajive.predict(return_dict=True)
for i in np.arange(100):
j = np.sum(blocks[0]["individual"][i] == blocks[1]["individual"][i])
assert j == 20
def test_ajive_plot(data):
x = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.fit(Xs=x)
blocks = ajive.predict(return_dict=True)
ajive_full_estimate_heatmaps(x, blocks)
p = 1
assert p == 1
def test_ajive_plot_list(data):
x = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.fit(Xs=x)
blocks = ajive.predict(return_dict=False)
ajive_full_estimate_heatmaps(x, blocks, names=["x1", "x2"])
p = 1
assert p == 1
def test_check_sparse(data):
dat = data["sparse_views"]
spar_mat = dat[0]
assert np.sum(spar_mat == 0) > np.sum(spar_mat != 0)
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.fit(Xs=dat)
blocks = ajive.predict(return_dict=True)
assert np.sum(np.sum(blocks[0]["individual"] == 0)) > np.sum(
np.sum(blocks[0]["individual"] != 0))
def test_wrong_sig(data):
dat = data["diff_views"]
ajive = AJIVE(init_signal_ranks=[-1, -4])
try:
ajive.fit(Xs=dat)
j = 0
except:
j = 1
assert j == 1
def test_precomp_init_svd(data):
dat = data["same_views"]
precomp = []
for i in dat:
precomp.append(svd_wrapper(i))
ajive = AJIVE(init_signal_ranks=[2, 2], joint_rank=1)
ajive.fit(dat, precomp_init_svd=precomp)
p = 3
assert p == 3
def test_dont_store_full(self):
"""
Make sure setting store_full = False works
"""
ajive = AJIVE(init_signal_ranks=[2, 3], store_full=False)
ajive.fit(Xs=[self.X, self.Y])
self.assertTrue(ajive.blocks_[0].joint.full_ is None)
self.assertTrue(ajive.blocks_[0].individual.full_ is None)
self.assertTrue(ajive.blocks_[1].joint.full_ is None)
self.assertTrue(ajive.blocks_[1].individual.full_ is None)
def setUp(self):
np.random.seed(12)
# First View
V1_joint = np.bmat([[-1 * np.ones((10, 20))], [np.ones((10, 20))]])
V1_joint = np.bmat([np.zeros((20, 80)), V1_joint])
V1_indiv_t = np.bmat([
[np.ones((4, 50))],
[-1 * np.ones((4, 50))],
[np.zeros((4, 50))],
[np.ones((4, 50))],
[-1 * np.ones((4, 50))],
])
V1_indiv_b = np.bmat([[np.ones((5, 50))], [-1 * np.ones((10, 50))],
[np.ones((5, 50))]])
V1_indiv_tot = np.bmat([V1_indiv_t, V1_indiv_b])
V1_noise = np.random.normal(loc=0, scale=1, size=(20, 100))
# Second View
V2_joint = np.bmat([[np.ones((10, 10))], [-1 * np.ones((10, 10))]])
V2_joint = 5000 * np.bmat([V2_joint, np.zeros((20, 10))])
V2_indiv = 5000 * np.bmat([
[-1 * np.ones((5, 20))],
[np.ones((5, 20))],
[-1 * np.ones((5, 20))],
[np.ones((5, 20))],
])
V2_noise = 5000 * np.random.normal(loc=0, scale=1, size=(20, 20))
# View Construction
X = V1_indiv_tot + V1_joint + V1_noise
Y = V2_indiv + V2_joint + V2_noise
obs_names = ["sample_{}".format(i) for i in range(X.shape[0])]
var_names = {
"x": ["x_var_{}".format(i) for i in range(X.shape[1])],
"y": ["y_var_{}".format(i) for i in range(Y.shape[1])],
}
X = pd.DataFrame(X, index=obs_names, columns=var_names["x"])
Y = pd.DataFrame(Y, index=obs_names, columns=var_names["y"])
self.ajive = AJIVE(init_signal_ranks=[2, 3]).fit(Xs=[X, Y],
view_names=["x", "y"])
self.X = X
self.Y = Y
self.obs_names = obs_names
self.var_names = var_names
def test_fit_elbows():
n = 10
elbows = 3
np.random.seed(1)
x = np.random.binomial(1, 0.6, (n**2)).reshape(n, n)
xorth = orth(x)
d = np.zeros(xorth.shape[0])
for i in range(0, len(d), int(len(d) / (elbows + 1))):
d[:i] += 10
X = xorth.T.dot(np.diag(d)).dot(xorth)
Xs = [X, X]
ajive = AJIVE(n_elbows=2)
ajive = ajive.fit(Xs)
np.testing.assert_equal(list(ajive.init_signal_ranks_.values())[0], 4)
def test_centering(self):
xmean = self.X.mean(axis=0)
ymean = self.Y.mean(axis=0)
self.assertTrue(np.allclose(self.ajive.centers_["x"], xmean))
self.assertTrue(np.allclose(self.ajive.blocks_["x"].joint.m_, xmean))
self.assertTrue(
np.allclose(self.ajive.blocks_["x"].individual.m_, xmean))
self.assertTrue(np.allclose(self.ajive.centers_["y"], ymean))
self.assertTrue(np.allclose(self.ajive.blocks_["y"].joint.m_, ymean))
self.assertTrue(
np.allclose(self.ajive.blocks_["y"].individual.m_, ymean))
# no centering
ajive = AJIVE(init_signal_ranks=[2, 3], center=False)
ajive = ajive.fit(Xs=[self.X, self.Y], view_names=["x", "y"])
self.assertTrue(ajive.centers_["x"] is None)
self.assertTrue(ajive.centers_["y"] is None)
# only center x
ajive = AJIVE(init_signal_ranks=[2, 3], center=[True, False])
ajive = ajive.fit(Xs=[self.X, self.Y], view_names=["x", "y"])
self.assertTrue(np.allclose(ajive.centers_["x"], xmean))
self.assertTrue(ajive.centers_["y"] is None)
def test_rank0(self):
"""
Check setting joint/individual rank to zero works
"""
ajive = AJIVE(init_signal_ranks=[2, 3], joint_rank=0)
ajive.fit(Xs=[self.X, self.Y])
self.assertTrue(ajive.common_.rank == 0)
self.assertTrue(ajive.blocks_[0].joint.rank == 0)
self.assertTrue(ajive.blocks_[0].joint.scores_ is None)
ajive = AJIVE(init_signal_ranks=[2, 3], indiv_ranks=[0, 1])
ajive.fit(Xs=[self.X, self.Y])
self.assertTrue(ajive.blocks_[0].individual.rank == 0)
self.assertTrue(ajive.blocks_[0].individual.scores_ is None)
def test_decomp_not_computed_ranks():
with pytest.raises(ValueError):
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.get_ranks()
def test_traditional_output(data):
x = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.fit(Xs=x, view_names=["x", "y"])
ajive.predict(return_dict=False)
def test_name_values_type(data):
with pytest.raises(ValueError):
x = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.fit(Xs=x, view_names={"jon": "first", "rich": "second"})
def test_name_values(data):
with pytest.raises(ValueError):
x = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.fit(Xs=x, view_names=["1", "2", "3"])
def test_n_wedin():
ajive = AJIVE(init_signal_ranks=[2, 2], n_wedin_samples=6)
assert ajive.n_wedin_samples == 6
def test_joint_noise_length(data):
dat = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.fit(Xs=dat)
blocks = ajive.predict(return_dict=True)
assert blocks[0]["joint"].shape == blocks[0]["noise"].shape
def test_is_fit():
ajive = AJIVE(init_signal_ranks=[2, 2], joint_rank=2)
assert ajive.is_fit_ == False
def test_n_randdir():
ajive = AJIVE(init_signal_ranks=[2, 2], n_randdir_samples=5)
assert ajive.n_randdir_samples == 5
def test_joint_rank(data):
dat = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2], joint_rank=2)
ajive.fit(Xs=dat)
assert ajive.joint_rank == 2
def test_indiv_rank(data):
dat = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2], indiv_ranks=[2, 1])
ajive.fit(Xs=dat)
assert ajive.indiv_ranks[0] == 2
def test_check_gen_lin_op_scipy(data):
with pytest.raises(TypeError):
dat = data["bad_views"]
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.fit(Xs=dat)
def test_check_joint_rank_large(data):
with pytest.raises(ValueError):
dat = data["same_views"]
ajive = AJIVE(init_signal_ranks=[2, 2], joint_rank=5)
ajive.fit(Xs=dat)
def test_get_ranks(data):
with pytest.raises(ValueError):
ajive = AJIVE(init_signal_ranks=[2, 2])
ajive.get_ranks()