def test_base_bmodel_indicator_not_inputs():
"""Test that we cannot pass as indicator nodes that are input of other nodes"""
J = np.array([[0, 1, 0], [0, 0, 1], [0, 0, 0]])
# node 1 cannot be indicator
with pytest.raises(IndicatorError):
_ = Bmodel(J=J, indicator_nodes=["node_1"])
# node 2 cannot be indicator
with pytest.raises(IndicatorError):
_ = Bmodel(J=J, indicator_nodes=["node_2"])
# instead node 0 can
_ = Bmodel(J=J, indicator_nodes=["node_0"])
assert True
def test_base_bmodel_indicator_not_inputs_using_labels():
"""Test that we cannot pass as indicator nodes that are input of other nodes, using string labels"""
J = np.array([[0, 1, 0], [0, 0, 1], [0, 0, 0]])
# node 1 cannot be indicator
with pytest.raises(IndicatorError):
_ = Bmodel(J=J, node_labels=["a", "b", "c"], indicator_nodes=["b"])
# node 2 cannot be indicator
with pytest.raises(IndicatorError):
_ = Bmodel(J=J, node_labels=["a", "b", "c"], indicator_nodes=["c"])
# instead node 0 can
_ = Bmodel(J=J, node_labels=["a", "b", "c"], indicator_nodes=["a"])
assert True
def test_base_bmodel_no_hold():
"""Test the no hold option"""
n_nodes = 12
n_runs_pert = 16
bmodel = Bmodel(J=random_interaction_matrix(N=n_nodes))
bmodel.runs(16)
hold = False
for config in bmodel.steady_states.values:
for node_to_switch in bmodel.node_labels:
for switch_to in ["ON", "OFF"]:
bmodel.perturbe(initial_condition=config,
node_to_switch=node_to_switch,
switch_to=switch_to,
n_runs=n_runs_pert,
hold=hold)
# if node is not holded, then final state can or cannot be what we asked for
# this is difficult to test
checks = []
for switched_node in bmodel.node_labels:
for switch_to in ["ON", "OFF"]:
perturbations = bmodel.get_perturbations(
switched_node=switched_node, switch_to=switch_to, hold=hold)
checks.append((perturbations[switched_node] == {
"ON": -1,
"OFF": 1
}[switch_to]).values)
checks = list(np.concatenate(checks))
checks = sorted(list(set(checks)))
assert checks == [False, True]
def test_base_bmodel_hold():
"""Test the hold option"""
n_nodes = 12
n_runs_pert = 4
bmodel = Bmodel(J=random_interaction_matrix(N=n_nodes))
bmodel.runs(16)
hold = True
for config in bmodel.steady_states.values:
for node_to_switch in bmodel.node_labels:
for switch_to in ["ON", "OFF"]:
bmodel.perturbe(initial_condition=config,
node_to_switch=node_to_switch,
switch_to=switch_to,
n_runs=n_runs_pert,
hold=hold)
# If node is holded, its final state must be what we asked for
for switched_node in bmodel.node_labels:
for switch_to in ["ON", "OFF"]:
perturbations = bmodel.get_perturbations(
switched_node=switched_node, switch_to=switch_to, hold=hold)
assert np.all(perturbations[switched_node] == {
"ON": 1,
"OFF": -1
}[switch_to])
def test_base_bmodel_get_perturbations():
"""Test that we retrieve the right number of perturbations"""
n_nodes = 16
n_runs_pert = 5
bmodel = Bmodel(J=random_interaction_matrix(N=n_nodes), maxT=4)
bmodel.runs(20)
# do the perturbations
for config in bmodel.steady_states.values:
for hold in [True, False]:
for node_to_switch in bmodel.node_labels:
for switch_to in ["ON", "OFF"]:
bmodel.perturbe(initial_condition=config,
node_to_switch=node_to_switch,
switch_to=switch_to,
n_runs=n_runs_pert,
hold=hold)
# count the perturbations
for hold in [True, False]:
for switched_node in bmodel.node_labels:
for switch_to in ["ON", "OFF"]:
perturbations = bmodel.get_perturbations(
switched_node=switched_node,
switch_to=switch_to,
hold=hold)
n_ics = (bmodel.steady_states[switched_node] == {
"ON": -1,
"OFF": 1
}[switch_to]).sum()
assert perturbations.shape == (n_ics * n_runs_pert, n_nodes)
def test_base_bmodel_indicator_can_fail():
n_nodes = 16
J = random_interaction_matrix(N=n_nodes)
node_labels = ["node_%d" % i for i in range(n_nodes)]
indicator_nodes = ["node_a", "node_b"]
with pytest.raises(IndicatorError):
_ = Bmodel(J=J,
node_labels=node_labels,
indicator_nodes=indicator_nodes)
def test_base_bmodel_indicator_unique():
"""Test that we cannot pass non-unique indicators"""
n_nodes = 16
J = random_interaction_matrix(N=n_nodes)
node_labels = ["node_%d" % i for i in range(n_nodes)]
indicator_nodes = ["node_0", "node_1", "node_0"]
with pytest.raises(IndicatorError):
_ = Bmodel(J=J,
node_labels=node_labels,
indicator_nodes=indicator_nodes)
def test_base_bmodel_indicator_idx():
"""Test that we correctly save the positions of indicator nodes"""
n_nodes = 4
J = np.array([
[0, 1, 0, 0],
[0, 0, 1, 0],
[0, 0, 0, 0],
[0, 1, 0, 0],
])
node_labels = ["node_%d" % i for i in range(n_nodes)]
indicator_nodes = ["node_0", "node_3"]
bmodel = Bmodel(J=J,
node_labels=node_labels,
indicator_nodes=indicator_nodes)
assert bmodel._indicator_idx == [0, 3]
def test_base_bmodel_indicator_is_set():
n_nodes = 16
node_labels = ["node_%d" % i for i in range(n_nodes)]
# let us get at least two indicator nodes
num_indicators = 0
while num_indicators <= 1:
J = random_interaction_matrix(N=n_nodes)
indicator_nodes = [
node for i, node in enumerate(node_labels) if np.all(J[:, i] == 0)
]
num_indicators = len(indicator_nodes)
bmodel = Bmodel(J=J,
node_labels=node_labels,
indicator_nodes=indicator_nodes)
assert bmodel.indicator_nodes == list(indicator_nodes)
def test_base__fill_in_indicator_nodes():
"""Test that we can fill in indicator nodes"""
J = np.array([[0, 1, 0], [0, 0, 1], [0, 0, 0]])
bmodel = Bmodel(J=J, node_labels=["a", "b", "c"], indicator_nodes=["a"])
# get some steady states
bmodel.runs(20)
# erase indicator nodes, fill in indicator nodes, and make sure nothing else changed
for t in range(len(bmodel.steady_states)):
initial_condition = bmodel.steady_states.values[t]
initial_condition[bmodel._indicator_idx] = 0
ic = bmodel._parse_initial_condition(initial_condition)
ic_filled = bmodel._fill_in_indicator_nodes(ic)
for i, (a, b) in enumerate(zip(ic, ic_filled)):
if i in bmodel._indicator_idx:
assert a == 0
assert b in [-1, 1]
else:
assert a == b
def test_base_bmodel_init():
"""Test that bmodel objects can be instantiated."""
J = np.array([[0, -1], [-1, 0]])
bmodel = Bmodel(J=J)
assert isinstance(bmodel, Bmodel)
def bmodel_abc_chain():
J = np.array([[0, 1, 0], [0, 0, 1], [0, 0, 0]])
bmodel = Bmodel(J=J, node_labels=["a", "b", "c"], indicator_nodes=["a"])
return bmodel
def bmodel_neg_feedback():
"""Create simple network for testing."""
J = np.array([[0, -1], [-1, 0]])
node_labels = ["node0", "node1"]
bmodel = Bmodel(J=J, node_labels=node_labels, maxT=100)
return bmodel