def test_matrices(self):
'''Test adjacency and Laplacian matrix computation.'''
# Load data from text file to compare with the load result
p = itu.read_pedigree_from_test_file(itu.HUTT_PED)
A = sym_adjacency(p.graph)
L = laplacian(p.graph)
assert not np.nonzero(A-A.T)[0], 'Adjacency matrix returned from sym_adjacency() should be symmetric'
assert not np.nonzero(L-L.T)[0], 'Laplacian matrix should be symmetric'
assert (abs(L.sum(axis=1)) < 1e-8).all(), 'Laplacian matrix should be zero-row-sum'
def test_matrices(self):
'''Test adjacency and Laplacian matrix computation.'''
# Load data from text file to compare with the load result
p = itu.read_pedigree_from_test_file(itu.HUTT_PED)
A = sym_adjacency(p.graph)
L = laplacian(p.graph)
assert not np.nonzero(
A - A.T
)[0], 'Adjacency matrix returned from sym_adjacency() should be symmetric'
assert not np.nonzero(L -
L.T)[0], 'Laplacian matrix should be symmetric'
assert (abs(L.sum(axis=1)) <
1e-8).all(), 'Laplacian matrix should be zero-row-sum'
def _layout_positions(g, g_extended=None):
'''Return a dictionary that maps pedigree nodes to graph layout positions (x,y). Uses
a layered spring model. Capable of handling both normal and extended pedigree graphs.'''
g_extended = g_extended if g_extended else g
# x-positions: spring model
L = laplacian(g_extended, dtype=np.float)
[_, v] = eigsh(L, 2, which='SM')
x = v[:, 1]
# Compute depths based on -original- graph
y = pt.node_depth(g)
for node in g_extended.nodes_iter():
# Marriage node: 0.5 above child depth. Note that marriage nodes must have children
if node < 0:
y[node] = y[g_extended.successors_iter(node).next()] - 0.5
ymax = max(d for d in y.itervalues())
# Reverse the y-axis (generation 0 on top)
return dict(zip(g_extended.nodes_iter(), zip(x, ((ymax - y[node]) for node in g_extended.nodes_iter()))))
def _layout_positions(g, g_extended=None):
'''Return a dictionary that maps pedigree nodes to graph layout positions (x,y). Uses
a layered spring model. Capable of handling both normal and extended pedigree graphs.'''
g_extended = g_extended if g_extended else g
# x-positions: spring model
L = laplacian(g_extended, dtype=np.float)
[_, v] = eigsh(L, 2, which='SM')
x = v[:, 1]
# Compute depths based on -original- graph
y = pt.node_depth(g)
for node in g_extended.nodes_iter():
# Marriage node: 0.5 above child depth. Note that marriage nodes must have children
if node < 0:
y[node] = y[g_extended.successors_iter(node).next()] - 0.5
ymax = max(d for d in y.itervalues())
# Reverse the y-axis (generation 0 on top)
return dict(
zip(g_extended.nodes_iter(),
zip(x, ((ymax - y[node]) for node in g_extended.nodes_iter()))))
