def convert_hhgate(gate):
"""Convert a MOOSE gate into GateHHRates in NeuroML"""
hh_rates = neuroml.GateHHRates(id=gate.id_.value, name=gate.name)
alpha = gate.tableA.copy()
beta = gate.tableB - alpha
vrange = np.linspace(gate.min, gate.max, len(alpha))
afn, ap = hhfit.find_ratefn(vrange, alpha)
bfn, bp = hhfit.find_ratefn(vrange, beta)
if afn is None:
raise Exception('could not find a fitting function for `alpha`')
if bfn is None:
raise Exception('could not find a fitting function for `alpha`')
afn_type = fn_rate_map[afn]
afn_component_type = None
if afn_type is None:
afn_type, afn_component_type = define_component_type(afn)
hh_rates.forward_rate = neuroml.HHRate(type=afn_type,
midpoint='%gmV' % (ap[2]),
scale='%gmV' % (ap[1]),
rate='%gper_ms' % (ap[0]))
bfn_type = fn_rate_map[bfn]
bfn_component_type = None
if bfn_type is None:
bfn_type, bfn_component_type = define_component_type(bfn)
hh_rates.reverse_rate = neuroml.HHRate(type=bfn_type,
midpoint='%gmV' % (bp[2]),
scale='%gmV' % (bp[1]),
rate='%gper_ms' % (bp[0]))
return hh_rates, afn_component_type, bfn_component_type
def convert_hhgate(gate):
"""Convert a MOOSE gate into GateHHRates in NeuroML"""
hh_rates = neuroml.GateHHRates(id=gate.id_.value, name=gate.name)
alpha = gate.tableA.copy()
beta = gate.tableB - alpha
vrange = np.linspace(gate.min, gate.max, len(alpha))
afn, ap = hhfit.find_ratefn(vrange, alpha)
bfn, bp = hhfit.find_ratefn(vrange, beta)
if afn is None:
raise Exception('could not find a fitting function for `alpha`')
if bfn is None:
raise Exception('could not find a fitting function for `alpha`')
afn_type = fn_rate_map[afn]
afn_component_type = None
if afn_type is None:
afn_type, afn_component_type = define_component_type(afn)
hh_rates.forward_rate = neuroml.HHRate(type=afn_type,
midpoint='%gmV' % (ap[2]),
scale='%gmV' % (ap[1]),
rate='%gper_ms' % (ap[0]))
bfn_type = fn_rate_map[bfn]
bfn_component_type = None
if bfn_type is None:
bfn_type, bfn_component_type = define_component_type(bfn)
hh_rates.reverse_rate = neuroml.HHRate(type=bfn_type,
midpoint='%gmV' % (bp[2]),
scale='%gmV' % (bp[1]),
rate='%gper_ms' % (bp[0]))
return hh_rates, afn_component_type, bfn_component_type
def test_sigmoid(self):
print('Testing sigmoid')
fn, params = hhfit.find_ratefn(self.v_array, self.sigmoid)
print('Sigmoid params original:', self.p_sigmoid, 'detected:', params)
pylab.plot(self.v_array, self.sigmoid, 'y-',
self.v_array, hhfit.sigmoid(self.v_array, *self.p_sigmoid), 'b--',
self.v_array, fn(self.v_array, *params), 'r-.')
pylab.legend(('original sigmoid', 'computed', 'fitted %s' % (fn)))
pylab.show()
self.assertEqual(hhfit.sigmoid, fn)
rms_error = np.sqrt(np.mean((self.sigmoid - fn(self.v_array, *params))**2))
self.assertAlmostEqual(rms_error/max(abs(self.sigmoid)), 0.0, places=3)
def test_sigmoid(self):
print('Testing sigmoid')
fn, params = hhfit.find_ratefn(self.v_array, self.sigmoid)
print('Sigmoid params original:', self.p_sigmoid, 'detected:', params)
pylab.plot(self.v_array, self.sigmoid, 'y-',
self.v_array, hhfit.sigmoid(self.v_array, *self.p_sigmoid), 'b--',
self.v_array, fn(self.v_array, *params), 'r-.')
pylab.legend(('original sigmoid', 'computed', 'fitted %s' % (fn)))
pylab.show()
self.assertEqual(hhfit.sigmoid, fn)
rms_error = np.sqrt(np.mean((self.sigmoid - fn(self.v_array, *params))**2))
self.assertAlmostEqual(rms_error/max(abs(self.sigmoid)), 0.0, places=3)
def test_dblexponential(self):
print('Testing double exponential')
fn, params = hhfit.find_ratefn(self.v_array, self.dblexp)
fnval = fn(self.v_array, *params)
pylab.plot(self.v_array, self.dblexp, 'y-',
self.v_array, hhfit.double_exp(self.v_array, *self.p_dblexp), 'b--',
self.v_array, fnval, 'r-.')
pylab.legend(('original dblexp', 'computed', 'fitted %s' % (fn)))
pylab.show()
self.assertEqual(hhfit.double_exp, fn)
rms_error = np.sqrt(np.mean((self.dblexp - fnval)**2))
print(params, rms_error)
self.assertAlmostEqual(rms_error/max(self.dblexp), 0.0, places=3)
def test_dblexponential(self):
print('Testing double exponential')
fn, params = hhfit.find_ratefn(self.v_array, self.dblexp)
fnval = fn(self.v_array, *params)
pylab.plot(self.v_array, self.dblexp, 'y-',
self.v_array, hhfit.double_exp(self.v_array, *self.p_dblexp), 'b--',
self.v_array, fnval, 'r-.')
pylab.legend(('original dblexp', 'computed', 'fitted %s' % (fn)))
pylab.show()
self.assertEqual(hhfit.double_exp, fn)
rms_error = np.sqrt(np.mean((self.dblexp - fnval)**2))
print(params, rms_error)
self.assertAlmostEqual(rms_error/max(self.dblexp), 0.0, places=3)
def test_linoid(self):
print 'Testing linoid'
fn, params = hhfit.find_ratefn(self.v_array, self.linoid)
print 'Linoid params original:', self.p_linoid, 'detected:', params
pylab.plot(self.v_array, self.linoid, 'y-',
self.v_array, hhfit.linoid(self.v_array, *self.p_linoid), 'b--',
self.v_array, fn(self.v_array, *params), 'r-.')
pylab.legend(('original linoid', 'computed', 'fitted %s' % (fn)))
pylab.show()
self.assertEqual(hhfit.linoid, fn)
fnval = fn(self.v_array, *params)
rms_error = np.sqrt(np.mean((self.linoid - fnval)**2))
self.assertAlmostEqual(rms_error/max(self.linoid), 0.0, places=3)
def test_exponential(self):
print('Testing exponential')
fn, params = hhfit.find_ratefn(self.v_array, self.exp)
print('Exponential params original:', self.p_exp, 'detected:', params)
fnval = hhfit.exponential(self.v_array, *params)
pylab.plot(self.v_array, self.exp, 'y-',
self.v_array, hhfit.exponential(self.v_array, *self.p_exp), 'b--',
self.v_array, fnval, 'r-.')
pylab.legend(('original exp', 'computed', 'fitted %s' % (fn)))
pylab.show()
self.assertEqual(hhfit.exponential, fn)
# The same exponential can be satisfied by an infinite number
# of parameter values. Hence we cannot compare the parameters,
# but only the fit
rms_error = np.sqrt(np.sum((self.exp - fnval)**2))
# pylab.plot(self.v_array, self.exp, 'b-')
# pylab.plot(self.v_array, fnval, 'r-.')
# pylab.show()
print(rms_error, rms_error/max(self.exp))
self.assertAlmostEqual(rms_error/max(self.exp), 0.0, places=3)
def test_exponential(self):
print('Testing exponential')
fn, params = hhfit.find_ratefn(self.v_array, self.exp)
print('Exponential params original:', self.p_exp, 'detected:', params)
fnval = hhfit.exponential(self.v_array, *params)
pylab.plot(self.v_array, self.exp, 'y-',
self.v_array, hhfit.exponential(self.v_array, *self.p_exp), 'b--',
self.v_array, fnval, 'r-.')
pylab.legend(('original exp', 'computed', 'fitted %s' % (fn)))
pylab.show()
self.assertEqual(hhfit.exponential, fn)
# The same exponential can be satisfied by an infinite number
# of parameter values. Hence we cannot compare the parameters,
# but only the fit
rms_error = np.sqrt(np.sum((self.exp - fnval)**2))
# pylab.plot(self.v_array, self.exp, 'b-')
# pylab.plot(self.v_array, fnval, 'r-.')
# pylab.show()
print(rms_error, rms_error/max(self.exp))
self.assertAlmostEqual(rms_error/max(self.exp), 0.0, places=3)
