def _initTree1(self):
# 1 2
# \ /
# 0
croot = CompartmentNode(0, loc_ind=0)
cnode1 = CompartmentNode(1, loc_ind=1)
cnode2 = CompartmentNode(2, loc_ind=2)
ctree = CompartmentTree(root=croot)
ctree.addNodeWithParent(cnode1, croot)
ctree.addNodeWithParent(cnode2, croot)
self.ctree = ctree
def _initTree2(self):
# 3
# |
# 2
# |
# 1
# |
# 0
croot = CompartmentNode(0, loc_ind=0)
cnode1 = CompartmentNode(1, loc_ind=1)
cnode2 = CompartmentNode(2, loc_ind=2)
cnode3 = CompartmentNode(3, loc_ind=3)
ctree = CompartmentTree(root=croot)
ctree.addNodeWithParent(cnode1, croot)
ctree.addNodeWithParent(cnode2, cnode1)
ctree.addNodeWithParent(cnode3, cnode2)
self.ctree = ctree
def _initTree3(self):
# 4 5 6 7 8
# \|/ \ /
# 1 2 3
# \ | /
# \|/
# 0
cns = [CompartmentNode(ii, loc_ind=ii) for ii in range(9)]
ctree = CompartmentTree(root=cns[0])
# first order children
ctree.addNodeWithParent(cns[1], cns[0])
ctree.addNodeWithParent(cns[2], cns[0])
ctree.addNodeWithParent(cns[3], cns[0])
# second order children
ctree.addNodeWithParent(cns[4], cns[1])
ctree.addNodeWithParent(cns[5], cns[1])
ctree.addNodeWithParent(cns[6], cns[1])
ctree.addNodeWithParent(cns[7], cns[3])
ctree.addNodeWithParent(cns[8], cns[3])
self.ctree = ctree
class TestReducedNeuron():
def addLocinds(self):
for ii, cn in enumerate(self.ctree):
cn.loc_ind = ii
def loadTwoCompartmentModel(self, w_locinds=True):
# simple two compartment model
pnode = CompartmentNode(0, ca=1.5e-5, g_l=2e-3)
self.ctree = CompartmentTree(root=pnode)
cnode = CompartmentNode(1, ca=2e-6, g_l=3e-4, g_c=4e-3)
self.ctree.addNodeWithParent(cnode, pnode)
if w_locinds:
self.addLocinds()
def loadTModel(self, w_locinds=True):
# simple T compartment model
# pnode = CompartmentNode(0, ca=1.5e-5, g_l=2e-3)
# self.ctree = CompartmentTree(root=pnode)
# cnode = CompartmentNode(1, ca=1.5e-6, g_l=2.5e-4, g_c=2e-3)
# self.ctree.addNodeWithParent(cnode, pnode)
# lnode0 = CompartmentNode(2, ca=1.5e-6, g_l=2.5e-4, g_c=2e-3)
# self.ctree.addNodeWithParent(lnode0, cnode)
# lnode1 = CompartmentNode(3, ca=1.5e-6, g_l=2.5e-4, g_c=2e-3)
# self.ctree.addNodeWithParent(lnode1, cnode)
pnode = CompartmentNode(0, ca=1.5e-5, g_l=2e-3)
self.ctree = CompartmentTree(root=pnode)
cnode = CompartmentNode(1, ca=2e-6, g_l=3e-4, g_c=4e-3)
self.ctree.addNodeWithParent(cnode, pnode)
lnode0 = CompartmentNode(2, ca=1.5e-6, g_l=2.5e-4, g_c=3e-3)
self.ctree.addNodeWithParent(lnode0, cnode)
lnode1 = CompartmentNode(3, ca=1.5e-6, g_l=2.5e-4, g_c=5e-3)
self.ctree.addNodeWithParent(lnode1, cnode)
if w_locinds:
self.addLocinds()
def loadThreeCompartmentModel(self, w_locinds=True):
# simple 3 compartment model
pnode = CompartmentNode(0, ca=1.9e-6, g_l=1.8e-3)
self.ctree = CompartmentTree(root=pnode)
cnode = CompartmentNode(1, ca=2.4e-6, g_l=0.3e-4, g_c=3.9)
self.ctree.addNodeWithParent(cnode, pnode)
lnode0 = CompartmentNode(2, ca=1.9e-6, g_l=0.3e-4, g_c=3.8e-3)
self.ctree.addNodeWithParent(lnode0, cnode)
if w_locinds:
self.addLocinds()
def loadMultiDendModel(self, w_locinds=True):
# simple 3 compartment model
pnode = CompartmentNode(0, ca=1.9e-6, g_l=1.8e-3)
self.ctree = CompartmentTree(root=pnode)
cnode0 = CompartmentNode(1, ca=2.4e-6, g_l=0.3e-4, g_c=3.9)
self.ctree.addNodeWithParent(cnode0, pnode)
cnode1 = CompartmentNode(2, ca=1.9e-6, g_l=0.4e-4, g_c=3.8e-3)
self.ctree.addNodeWithParent(cnode1, pnode)
cnode2 = CompartmentNode(3, ca=1.3e-6, g_l=0.5e-4, g_c=2.7e-2)
self.ctree.addNodeWithParent(cnode2, pnode)
if w_locinds:
self.addLocinds()
def testGeometry1(self):
fake_c_m = 1.
fake_r_a = 100. * 1e-6
factor_r_a = 1e-6
## test method 1
# test with two compartments
self.loadTwoCompartmentModel()
ctree = self.ctree
# check if fake geometry is correct
points, _ = ctree.computeFakeGeometry(fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
factor_r_a=1e-6,
delta=1e-14,
method=1)
# create a neuron comparemtns
comps = []
for ii, node in enumerate(ctree):
comps.append(h.Section())
h.pt3dadd(*points[ii][0], sec=comps[-1])
h.pt3dadd(*points[ii][1], sec=comps[-1])
h.pt3dadd(*points[ii][2], sec=comps[-1])
h.pt3dadd(*points[ii][3], sec=comps[-1])
comps[-1].Ra = fake_r_a * 1e6
comps[-1].cm = fake_c_m
comps[-1].nseg = 1
# check areas
assert np.abs(comps[0](0.5).area() * 1e-8 * fake_c_m -
ctree[0].ca) < 1e-12
assert np.abs(comps[1](0.5).area() * 1e-8 * fake_c_m -
ctree[1].ca) < 1e-12
# check whether resistances are correct
assert np.abs(comps[0](0.5).ri() - 1.) < 1e-6
assert np.abs((comps[1](0.5).ri() - 1. / ctree[1].g_c) /
comps[1](0.5).ri()) < 1e-6
assert np.abs((comps[1](0.5).ri() * factor_r_a - comps[1](1.).ri()) /
comps[1](1.).ri()) < 1e-6
# test with three compartments
self.loadThreeCompartmentModel()
ctree = self.ctree
# check if fake geometry is correct
points, _ = ctree.computeFakeGeometry(fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
factor_r_a=1e-6,
delta=1e-14,
method=1)
# create a neuron comparemtns
comps = []
for ii, node in enumerate(ctree):
comps.append(h.Section())
h.pt3dadd(*points[ii][0], sec=comps[-1])
h.pt3dadd(*points[ii][1], sec=comps[-1])
h.pt3dadd(*points[ii][2], sec=comps[-1])
h.pt3dadd(*points[ii][3], sec=comps[-1])
comps[-1].Ra = fake_r_a * 1e6
comps[-1].cm = fake_c_m
comps[-1].nseg = 1
# check areas
assert np.abs(comps[0](0.5).area() * 1e-8 * fake_c_m -
ctree[0].ca) < 1e-12
assert np.abs(comps[1](0.5).area() * 1e-8 * fake_c_m -
ctree[1].ca) < 1e-12
assert np.abs(comps[2](0.5).area() * 1e-8 * fake_c_m -
ctree[2].ca) < 1e-12
# check whether resistances are correct
assert np.abs(comps[0](0.5).ri() - 1.) < 1e-6
assert np.abs((comps[1](0.5).ri() - 1. / ctree[1].g_c) /
comps[1](0.5).ri()) < 1e-6
assert np.abs((comps[2](0.5).ri() - 1. / ctree[2].g_c) /
comps[2](0.5).ri()) < 1e-6
assert np.abs((comps[1](0.5).ri() * factor_r_a - comps[1](1.).ri()) /
comps[1](1.).ri()) < 1e-6
assert np.abs((comps[2](0.5).ri() * factor_r_a - comps[2](1.).ri()) /
comps[2](1.).ri()) < 1e-6
# test the T model
self.loadTModel()
ctree = self.ctree
# check if fake geometry is correct
points, _ = ctree.computeFakeGeometry(fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
factor_r_a=1e-6,
delta=1e-14,
method=1)
# create a neuron comparemtns
comps = []
for ii, node in enumerate(ctree):
comps.append(h.Section())
h.pt3dadd(*points[ii][0], sec=comps[-1])
h.pt3dadd(*points[ii][1], sec=comps[-1])
h.pt3dadd(*points[ii][2], sec=comps[-1])
h.pt3dadd(*points[ii][3], sec=comps[-1])
comps[-1].Ra = fake_r_a * 1e6
comps[-1].cm = fake_c_m
comps[-1].nseg = 1
# check areas
assert np.abs(comps[0](0.5).area() * 1e-8 * fake_c_m -
ctree[0].ca) < 1e-12
assert np.abs(comps[1](0.5).area() * 1e-8 * fake_c_m -
ctree[1].ca) < 1e-12
assert np.abs(comps[2](0.5).area() * 1e-8 * fake_c_m -
ctree[2].ca) < 1e-12
assert np.abs(comps[3](0.5).area() * 1e-8 * fake_c_m -
ctree[3].ca) < 1e-12
# check whether resistances are correct
assert np.abs(comps[0](0.5).ri() - 1.) < 1e-6
assert np.abs((comps[1](0.5).ri() - 1. / ctree[1].g_c) /
comps[1](0.5).ri()) < 1e-6
assert np.abs((comps[2](0.5).ri() - 1. / ctree[2].g_c) /
comps[2](0.5).ri()) < 1e-6
assert np.abs((comps[3](0.5).ri() - 1. / ctree[3].g_c) /
comps[3](0.5).ri()) < 1e-6
assert np.abs((comps[1](0.5).ri() * factor_r_a - comps[1](1.).ri()) /
comps[1](1.).ri()) < 1e-6
assert np.abs((comps[2](0.5).ri() * factor_r_a - comps[2](1.).ri()) /
comps[2](1.).ri()) < 1e-6
assert np.abs((comps[3](0.5).ri() * factor_r_a - comps[3](1.).ri()) /
comps[3](1.).ri()) < 1e-6
def testImpedanceProperties1(self):
fake_c_m = 1.
fake_r_a = 100. * 1e-6
# create the two compartment model without locinds
self.loadTwoCompartmentModel(w_locinds=False)
ctree = self.ctree
# check if error is raised if loc_inds have not been set
with pytest.raises(AttributeError):
ctree.calcImpedanceMatrix()
# create the two compartment model with locinds
self.loadTwoCompartmentModel()
ctree = self.ctree
# compute the impedance matrix exactly
z_mat_comp = ctree.calcImpedanceMatrix()
# create a neuron model
sim_tree = createReducedNeuronModel(ctree,
fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
method=1)
z_mat_sim = sim_tree.calcImpedanceMatrix([(0, 0.5), (1, 0.5)])
# create the three compartmental model
self.loadThreeCompartmentModel()
ctree = self.ctree
# compute the impedance matrix exactly
z_mat_comp = ctree.calcImpedanceMatrix()
# create a neuron model
sim_tree = createReducedNeuronModel(ctree,
fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
method=1)
z_mat_sim = sim_tree.calcImpedanceMatrix([(0, 0.5), (1, 0.5),
(2, 0.5)])
# create the T compartmental model
self.loadTModel()
ctree = self.ctree
# compute the impedance matrix exactly
z_mat_comp = ctree.calcImpedanceMatrix()
# create a neuron model
sim_tree = createReducedNeuronModel(ctree,
fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
method=1)
z_mat_sim = sim_tree.calcImpedanceMatrix([(0, 0.5), (1, 0.5), (2, 0.5),
(3, 0.5)])
assert np.allclose(z_mat_sim, z_mat_comp)
# create the multidend model
self.loadMultiDendModel()
ctree = self.ctree
# compute the impedance matrix exactly
z_mat_comp = ctree.calcImpedanceMatrix()
# create a neuron model
sim_tree = createReducedNeuronModel(ctree,
fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
method=1)
z_mat_sim = sim_tree.calcImpedanceMatrix([(0, 0.5), (1, 0.5), (2, 0.5),
(3, 0.5)])
assert np.allclose(z_mat_sim, z_mat_comp)
def testGeometry2(self):
fake_c_m = 1.
fake_r_a = 100. * 1e-6
factor_r_a = 1e-6
## test method 2
# test with two compartments
self.loadTwoCompartmentModel()
ctree = self.ctree
# check if fake geometry is correct
lengths, radii = ctree.computeFakeGeometry(fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
factor_r_a=1e-6,
delta=1e-14,
method=2)
# create a neuron comparemtns
comps = []
for ii, node in enumerate(ctree):
comps.append(h.Section())
comps[-1].diam = 2. * radii[ii] * 1e4
comps[-1].L = lengths[ii] * 1e4
comps[-1].Ra = fake_r_a * 1e6
comps[-1].cm = fake_c_m
comps[-1].nseg = 1
# check areas
assert np.abs(comps[0](0.5).area() * 1e-8 * fake_c_m -
ctree[0].ca) < 1e-12
assert np.abs(comps[1](0.5).area() * 1e-8 * fake_c_m -
ctree[1].ca) < 1e-12
# check whether resistances are correct
assert np.abs((comps[1](0.5).ri() - 1. / ctree[1].g_c) /
comps[1](0.5).ri()) < 1e-12
assert np.abs((comps[1](0.5).ri() - comps[1](1.).ri()) /
comps[1](1.).ri()) < 1e-12
# test with three compartments
self.loadThreeCompartmentModel()
ctree = self.ctree
# check if fake geometry is correct
lengths, radii = ctree.computeFakeGeometry(fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
factor_r_a=1e-6,
delta=1e-14,
method=2)
# create a neuron comparemtns
comps = []
for ii, node in enumerate(ctree):
comps.append(h.Section())
comps[-1].diam = 2. * radii[ii] * 1e4
comps[-1].L = lengths[ii] * 1e4
comps[-1].Ra = fake_r_a * 1e6
comps[-1].cm = fake_c_m
comps[-1].nseg = 1
# check areas
assert np.abs(comps[0](0.5).area() * 1e-8 * fake_c_m -
ctree[0].ca) < 1e-12
assert np.abs(comps[1](0.5).area() * 1e-8 * fake_c_m -
ctree[1].ca) < 1e-12
assert np.abs(comps[2](0.5).area() * 1e-8 * fake_c_m -
ctree[2].ca) < 1e-12
# check whether resistances are correct
assert np.abs((comps[1](0.5).ri() - 1. / ctree[1].g_c) /
comps[1](0.5).ri()) < 1e-12
assert np.abs((comps[1](0.5).ri() - comps[1](1.).ri()) /
comps[1](1.).ri()) < 1e-12
assert np.abs((comps[2](0.5).ri() - 1. / ctree[2].g_c) /
comps[2](0.5).ri()) < 1e-12
assert np.abs((comps[2](0.5).ri() - comps[2](1.).ri()) /
comps[2](1.).ri()) < 1e-12
# test the T model
self.loadTModel()
ctree = self.ctree
# check if fake geometry is correct
lengths, radii = ctree.computeFakeGeometry(fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
factor_r_a=1e-6,
delta=1e-14,
method=2)
# create a neuron comparemtns
comps = []
for ii, node in enumerate(ctree):
comps.append(h.Section())
comps[-1].diam = 2. * radii[ii] * 1e4
comps[-1].L = lengths[ii] * 1e4
comps[-1].Ra = fake_r_a * 1e6
comps[-1].cm = fake_c_m
comps[-1].nseg = 1
# check areas
assert np.abs(comps[0](0.5).area() * 1e-8 * fake_c_m -
ctree[0].ca) < 1e-12
assert np.abs(comps[1](0.5).area() * 1e-8 * fake_c_m -
ctree[1].ca) < 1e-12
assert np.abs(comps[2](0.5).area() * 1e-8 * fake_c_m -
ctree[2].ca) < 1e-12
assert np.abs(comps[3](0.5).area() * 1e-8 * fake_c_m -
ctree[3].ca) < 1e-12
# check whether resistances are correct
assert np.abs((comps[1](0.5).ri() - 1. / ctree[1].g_c) /
comps[1](0.5).ri()) < 1e-12
assert np.abs((comps[1](0.5).ri() - comps[1](1.).ri()) /
comps[1](1.).ri()) < 1e-12
assert np.abs((comps[2](0.5).ri() - 1. / ctree[2].g_c) /
comps[2](0.5).ri()) < 1e-12
assert np.abs((comps[2](0.5).ri() - comps[2](1.).ri()) /
comps[2](1.).ri()) < 1e-12
assert np.abs((comps[3](0.5).ri() - 1. / ctree[3].g_c) /
comps[3](0.5).ri()) < 1e-12
assert np.abs((comps[3](0.5).ri() - comps[3](1.).ri()) /
comps[3](1.).ri()) < 1e-12
def testImpedanceProperties2(self):
fake_c_m = 1.
fake_r_a = 100. * 1e-6
# create the two compartment model
self.loadTwoCompartmentModel()
ctree = self.ctree
# compute the impedance matrix exactly
z_mat_comp = ctree.calcImpedanceMatrix()
# create a neuron model
sim_tree = createReducedNeuronModel(ctree,
fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
method=2)
z_mat_sim = sim_tree.calcImpedanceMatrix([(0, 0.5), (1, 0.5)])
assert np.allclose(z_mat_sim, z_mat_comp, atol=1e-2)
# create the three compartmental model
self.loadThreeCompartmentModel()
ctree = self.ctree
# compute the impedance matrix exactly
z_mat_comp = ctree.calcImpedanceMatrix()
# create a neuron model
sim_tree = createReducedNeuronModel(ctree,
fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
method=2)
z_mat_sim = sim_tree.calcImpedanceMatrix([(0, 0.5), (1, 0.5),
(2, 0.5)])
assert np.allclose(z_mat_sim, z_mat_comp)
# create the T compartmental model
self.loadTModel()
ctree = self.ctree
# compute the impedance matrix exactly
z_mat_comp = ctree.calcImpedanceMatrix()
# create a neuron model
sim_tree = createReducedNeuronModel(ctree,
fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
method=2)
z_mat_sim = sim_tree.calcImpedanceMatrix([(0, 0.5), (1, 0.5), (2, 0.5),
(3, 0.5)])
assert np.allclose(z_mat_sim, z_mat_comp)
# create the multidend model
self.loadMultiDendModel()
ctree = self.ctree
# compute the impedance matrix exactly
z_mat_comp = ctree.calcImpedanceMatrix()
# create a neuron model
sim_tree = createReducedNeuronModel(ctree,
fake_c_m=fake_c_m,
fake_r_a=fake_r_a,
method=2)
z_mat_sim = sim_tree.calcImpedanceMatrix([(0, 0.5), (1, 0.5), (2, 0.5),
(3, 0.5)])
assert np.allclose(z_mat_sim, z_mat_comp)