def test_model_check_data():
# Create a set of statements
a = Agent('A', db_refs={'HGNC': '1'})
b = Agent('B', db_refs={'HGNC': '2'})
b_phos = Agent('B',
mods=[ModCondition('phosphorylation')],
db_refs={'HGNC': '2'})
c = Agent('C', db_refs={'HGNC': '3'})
c_phos = Agent('C',
mods=[ModCondition('phosphorylation')],
db_refs={'HGNC': '3'})
d = Agent('D', db_refs={'HGNC': '4'})
d_phos = Agent('D',
mods=[ModCondition('phosphorylation')],
db_refs={'HGNC': '4'})
# Two paths from A to D: One going through B and another through C
st1 = Phosphorylation(a, b)
st2 = Phosphorylation(b_phos, d)
st3 = Phosphorylation(a, c)
st4 = Phosphorylation(c_phos, d)
# Statements/Data agents for checking
stmt_to_check = Phosphorylation(a, d)
agent_obs = [b_phos, c_phos, d_phos]
# Make model
pa = PysbAssembler()
pa.add_statements([st1, st2, st3, st4])
pa.make_model(policies='one_step')
mc = ModelChecker(pa.model, [stmt_to_check], agent_obs)
results = mc.check_model(max_paths=5)
mc.get_im().draw('im.pdf', prog='dot')
# Create observable
assert len(results) == 1
pr = results[0][1]
res = pr.paths[0:2]
assert len(res) == 2
p1 = [('A_phosphorylation_B_phospho', 1),
('B_phospho_phosphorylation_D_phospho', 1), ('D_phospho_p_obs', 1)]
assert p1 in res
p2 = [('A_phosphorylation_C_phospho', 1),
('C_phospho_phosphorylation_D_phospho', 1), ('D_phospho_p_obs', 1)]
assert p2 in res
# Now, a vector linking agents with values, expressed at first as
# +/- 1
# This data should ensure that the path through B should be more highly
# ranked than the path through C
data = {b_phos: 1, c_phos: -1, d_phos: 1}
paths = results[0][1].paths
scored_paths = mc.score_paths(paths, data)
assert scored_paths[0][0] == p1
assert scored_paths[0][1] == 0
assert scored_paths[1][0] == p2
def test_prune_influence_map():
kin = Agent('Kin', db_refs={'HGNC': '1'})
phos = Agent('Phos', db_refs={'HGNC': '2'})
subs = Agent('Substrate', db_refs={'HGNC': '3'})
st1 = Phosphorylation(kin, subs)
st2 = Dephosphorylation(phos, subs)
pa = PysbAssembler()
pa.add_statements([st1, st2])
pa.make_model(policies='one_step')
mc = ModelChecker(pa.model, [st1])
mc.prune_influence_map()
im = mc.get_im()
assert len(im.nodes()) == 3
assert len(im.edges()) == 2
