def add_edge_weights(
self,
potential_graph: nx.DiGraph(),
weights: Optional[Dict[int, Dict[int, str]]] = None,
missing_state_indicator: int = -1,
) -> nx.DiGraph:
"""Annotates edges with the weight.
Given a graph where nodes are iterable entities containing character
states, annotated edges with respect to the number of mutations. If a
prior dictionary is passed into the constructor, the log likelihood
of the mutations is added instead. These values are stored in the
`weight` attribute of the networkx graph.
Args:
potential_graph: Potential graph
weights: Weights to use when comparing states between characters
missing_state_indicator: Variable to indicate missing state
information.
Returns:
The potential graph with edge weights added, stored in the `weight`
attribute.
"""
weighted_graph = potential_graph.copy()
for u, v in weighted_graph.edges():
weighted_graph[u][v][
"weight"] = dissimilarity_functions.weighted_hamming_distance(
list(u), list(v), missing_state_indicator, weights)
return weighted_graph
def test_weighted_hamming_distance_priors_sq_inverse(self):
dissimilarity = dissimilarity_functions.weighted_hamming_distance(
self.s1, self.s2, weights=self.sqiweights)
expected_dissimilarity = np.sum([
np.sqrt(1 / self.priors[0][1]),
np.sqrt(1 / self.priors[5][2]) + np.sqrt(1 / self.priors[5][3]),
])
self.assertEqual(dissimilarity, expected_dissimilarity / 5)
def test_weighted_hamming_distance_priors_negative_log(self):
dissimilarity = dissimilarity_functions.weighted_hamming_distance(
self.s1, self.s2, weights=self.nlweights)
expected_dissimilarity = np.sum([
-np.log(self.priors[0][1]),
-(np.log(self.priors[5][2]) + np.log(self.priors[5][3])),
])
self.assertEqual(dissimilarity, expected_dissimilarity / 5)
def test_weighted_hamming_distance_all_missing(self):
dissimilarity = dissimilarity_functions.weighted_hamming_distance(
self.s1, self.all_missing, weights=self.nlweights)
self.assertEqual(dissimilarity, 0)
def test_weighted_hamming_distance_no_priors(self):
dissimilarity = dissimilarity_functions.weighted_hamming_distance(
self.s1, self.s2)
self.assertEqual(dissimilarity, 3 / 5)
def test_weighted_hamming_distance_identical(self):
dissimilarity = dissimilarity_functions.weighted_hamming_distance(
self.s1, self.s1)
self.assertEqual(dissimilarity, 0)