def simulate_example_inh_current(I_vec):
simTime = 1000. # ms
my_nest.ResetKernel()
model_list, model_dict = models()
my_nest.MyLoadModels(model_list, NEURON_MODELS)
df = my_nest.GetDefaults(NEURON_MODELS[0])
n = len(I_vec)
STN = MyGroup(NEURON_MODELS[0],
n,
sd=True,
mm=True,
mm_dt=1.0,
record_from=['V_m', 'u'])
I_e0 = my_nest.GetStatus(STN[:])[0]['I_e']
my_nest.SetStatus(STN[:], params={'I_e': I_e0 + I_E + 50}) # Set I_e
I_e = my_nest.GetStatus(STN.ids, 'I_e')[0]
scg = my_nest.Create('step_current_generator', n=n)
rec = my_nest.GetStatus(STN[:])[0]['receptor_types']
for source, target, I in zip(scg, STN[:], I_vec):
my_nest.SetStatus([source], {
'amplitude_times': [280., 700.],
'amplitude_values': [float(I), 0.]
})
my_nest.Connect([source], [target],
params={'receptor_type': rec['CURR']})
my_nest.MySimulate(simTime)
STN.get_signal('v', 'V_m', stop=simTime) # retrieve signal
STN.get_signal('s') # retrieve signal
STN.signals['V_m'].my_set_spike_peak(21, spkSignal=STN.signals['spikes'])
e = my_nest.GetStatus(STN.mm)[0]['events'] # get events
#pylab.plot(e['u'])
#pylab.show()
meanRate = round(STN.signals['spikes'].mean_rate(0, 500), 1)
s = '\n'
s = s + 'Example inhibitory current:\n'
s = s + ' %s %5s %3s %s %5s %3s \n' % ('Mean rate:', meanRate, 'Hz', 'I_e',
I_e, 'pA')
s = s + 'Steps:\n'
s = s + ' %5s %3s \n' % (I_vec, 'pA')
infoString = s
return STN, infoString
def simulate_ahp(I_vec):
simTime = 3000. # ms
my_nest.ResetKernel()
model_list, model_dict = models()
my_nest.MyLoadModels(model_list, NEURON_MODELS)
n = len(I_vec)
STN = MyGroup(NEURON_MODELS[0], n, sd=True, mm=True, mm_dt=1.0)
I_e0 = my_nest.GetStatus(STN[:])[0]['I_e']
#my_nest.SetStatus(STN[:], params={'I_e':-10.}) # Set I_e
my_nest.SetStatus(STN[:], params={'I_e': 1.0}) # Set I_e
I_e = my_nest.GetStatus(STN.ids, 'I_e')[0]
scg = my_nest.Create('step_current_generator', n=n)
rec = my_nest.GetStatus(STN[:])[0]['receptor_types']
for source, target, I in zip(scg, STN[:], I_vec):
my_nest.SetStatus([source], {
'amplitude_times': [500., 1000.],
'amplitude_values': [float(I), 0.]
})
my_nest.Connect([source], [target],
params={'receptor_type': rec['CURR']})
my_nest.MySimulate(simTime)
STN.get_signal('s') # retrieve signal
STN.signals['spikes'] = STN.signals['spikes'].time_slice(700, 2000)
delays = []
for i, curr in enumerate(I_vec):
delays.append(
max(
numpy.diff(
STN.signals['spikes'].spiketrains[i + 1.0].spike_times)))
meanRate = round(STN.signals['spikes'].mean_rate(0, 500), 1)
s = '\n'
s = s + 'Example inhibitory current:\n'
s = s + ' %s %5s %3s %s %5s %3s \n' % ('Mean rate:', meanRate, 'Hz', 'I_e',
I_e, 'pA')
s = s + 'Steps:\n'
s = s + ' %5s %3s \n' % (I_vec, 'pA')
infoString = s
return I_vec, delays
def simulate_IF(I_vec):
tStim = 700
my_nest.ResetKernel()
model_list, model_dict = models()
my_nest.MyLoadModels(model_list, NEURON_MODELS)
SNR = MyGroup(NEURON_MODELS[0], 1, mm=True, sd=True, mm_dt=0.1)
I_e0 = my_nest.GetStatus(SNR[:])[0]['I_e']
my_nest.SetStatus(SNR[:], params={'I_e': I_e0 + I_E}) # Set I_e
I_e = my_nest.GetStatus(SNR.ids, 'I_e')[0]
I_vec, fIsi, mIsi, lIsi = SNR.IF(I_vec, tStim=tStim)
speed_f = numpy.diff(1000.0 / fIsi) / numpy.diff(I_vec)
speed_l = numpy.diff(1000.0 / lIsi) / numpy.diff(I_vec)
speed_f = speed_f[speed_f > 0]
speed_l = speed_l[speed_l > 0]
s = '\n'
s = s + 'IF:\n'
s = s + ' %s %5s %3s \n' % ('First to Last ISI:', tStim, 'ms')
s = s + ' %s %5s %3s \n' % ('Added I_e:', I_e, 'pA')
s = s + ' %s %4s %s %4s %s %4s\n' % (
'Speed first ((Hz/pA), min:', str(min(speed_f))[0:4], 'max',
str(max(speed_f))[0:4], 'mean', str(sum(speed_f) / len(speed_f))[0:4])
s = s + ' %s %4s %s %4s %s %4s\n' % (
'Speed last (Hz/pA), min:', str(min(speed_l))[0:4], 'max',
str(max(speed_l))[0:4], 'mean', str(sum(speed_l) / len(speed_l))[0:4])
infoString = s
return I_vec, fIsi, mIsi, lIsi, infoString
def simulate_IV(I_vec):
my_nest.ResetKernel()
model_list, model_dict = models()
my_nest.MyLoadModels(model_list, NEURON_MODELS)
SNR = MyGroup(NEURON_MODELS[0], 1, sd=True, mm=True, mm_dt=0.1)
I_e0 = my_nest.GetStatus(SNR[:])[0]['I_e']
my_nest.SetStatus(SNR[:], params={'I_e': I_e0 + I_E}) # Set I_e
I_e = my_nest.GetStatus(SNR.ids, 'I_e')[0]
I_vec, voltage = SNR.IV_I_clamp(I_vec)
#current=current
speed = numpy.diff(voltage) / numpy.diff(I_vec) * 1000.
speed = speed[speed > 0]
s = '\n'
s = s + 'IV:\n'
s = s + ' %s %5s %3s \n' % ('I_e:', I_e, 'pA')
'''
s = s + ' %s %4s %s %4s %s %4s\n' % ( 'Speed (mV/pA=MOhm), min:',
str(min(speed))[0:4],
'max',str(max(speed))[0:4],
'mean',
str(sum(speed)/len(speed))[0:4])
'''#
infoString = s
I_vec = numpy.array(I_vec)
voltage = numpy.array(voltage)
return I_vec, voltage, infoString
def simulate_example_rebound_spike(I_vec):
simTime = 5000. # ms
my_nest.ResetKernel()
model_list, model_dict=models()
my_nest.MyLoadModels( model_list, NEURON_MODELS )
n=len(I_vec)
GPE = MyGroup( NEURON_MODELS[0], n, sd=True, mm=True, mm_dt = 1.0 )
I_e0=my_nest.GetStatus(GPE[:])[0]['I_e']
#my_nest.SetStatus(GPE[:], params={'I_e':-10.}) # Set I_e
my_nest.SetStatus(GPE[:], params={'I_e':-10.}) # Set I_e
I_e = my_nest.GetStatus(GPE.ids,'I_e')[0]
scg = my_nest.Create( 'step_current_generator',n=n )
rec=my_nest.GetStatus(GPE[:])[0]['receptor_types']
for source, target, I in zip(scg, GPE[:], I_vec):
my_nest.SetStatus([source], {'amplitude_times':[500.,700.],
'amplitude_values':[float(I),0.]})
my_nest.Connect( [source], [target],
params = { 'receptor_type' : rec['CURR'] } )
my_nest.MySimulate(simTime)
GPE.get_signal( 'v','V_m', stop=simTime ) # retrieve signal
GPE.get_signal( 's') # retrieve signal
GPE.signals['V_m'].my_set_spike_peak( 21, spkSignal= GPE.signals['spikes'] )
meanRate=round(GPE.signals['spikes'].mean_rate(0,500),1)
s='\n'
s =s + 'Example inhibitory current:\n'
s = s + ' %s %5s %3s %s %5s %3s \n' % ( 'Mean rate:', meanRate, 'Hz',
'I_e', I_e,'pA' )
s = s + 'Steps:\n'
s = s + ' %5s %3s \n' % ( I_vec, 'pA' )
infoString=s
return GPE, infoString
def simulate_voltage_ipsp(I_vec):
simTime = 700. # ms
spikes_at = numpy.arange(500., len(I_vec) * simTime, simTime) # ms
voltage = [] # mV
ipsp_weak = [] # mV
ipsp_strong = [] # mV
my_nest.ResetKernel()
model_list = models()
my_nest.MyLoadModels(model_list, NEURON_MODELS)
my_nest.MyLoadModels(model_list, SYNAPES_MODELS)
SNR = MyGroup(NEURON_MODELS[0], len(SYNAPES_MODELS), mm=True, mm_dt=0.1)
SG = my_nest.Create('spike_generator', params={'spike_times': spikes_at})
for i in range(len(SYNAPES_MODELS)):
my_nest.Connect(SG, [SNR[i]], model=SYNAPES_MODELS[i])
simTimeTot = 0
for I_e in I_vec:
my_nest.SetStatus(SNR[:], params={'I_e': float(I_e)})
my_nest.MySimulate(simTime)
simTimeTot += simTime
SNR.get_signal('v', 'V_m', stop=simTimeTot) # retrieve signal
simTimeAcum = 0
for I_e in I_vec:
signal = SNR.signals['V_m'].my_time_slice(400 + simTimeAcum,
700 + simTimeAcum)
simTimeAcum += simTime
clamped_at = signal[1].signal[-1]
minV = min(signal[1].signal)
maxV = max(signal[1].signal)
if abs(minV - clamped_at) < abs(maxV - clamped_at):
size_weak = max(signal[1].signal) - clamped_at
size_strong = max(signal[2].signal) - clamped_at
else:
size_weak = min(signal[1].signal) - clamped_at
size_strong = min(signal[2].signal) - clamped_at
voltage.append(clamped_at)
ipsp_weak.append(size_weak)
ipsp_strong.append(size_strong)
ipsp = numpy.array([ipsp_weak, ipsp_strong])
return voltage, ipsp
def create_output_population(nOutput, outputAddCurrent, outputName, sname,
spath):
create_models(outputName)
Output = MyGroup(outputName, nOutput, mm_dt=1.0, sname=sname, spath=spath)
I_e = my_nest.GetStatus(Output.local_ids, 'I_e')[0]
# add output current
my_nest.SetStatus(Output[:], {'I_e': I_e + outputAddCurrent})
return Output
def create_input_population(nInput, nRep=1, simTime=20000):
''' Define input as inhomogenous poisson processes '''
spike_times = []
inputName = 'spike_generator'
Input = MyGroup(inputName,
nInput,
mm_dt=1.0,
spath=spath,
sname='MSN',
mm=False,
sd=False)
for i in range(nInput):
rates = meanInputRates[:]
times = meanInputTimings[:]
if Input.ids[i] in selectedInputIds:
rates.extend(selectedInputRates)
times.extend(selectedInputTimings)
rates.append(inbetweenInputRate)
times.append(inbetweenInputTime)
spikes = misc.inh_poisson_spikes(rates,
times,
t_stop=simTime,
n_rep=nRep,
seed=i)
# create spike list for input
for spk in spikes:
spike_times.append((i, spk))
my_nest.SetStatus([Input.ids[i]], params={'spike_times': spikes})
# add spike list for input to input spike list
Input.signals['spikes'] = my_signals.MySpikeList(spike_times, Input.ids)
return Input
def simulate_network_poisson(params_msn_d1,
params_msn_d2,
params_stn,
synapse_models,
sim_time,
seed,
I_e_add,
threads=1,
start_rec=0,
model_params={},
params_in={},
p_weights=False,
p_conn=False,
p_I_e=False):
'''
Assume that the background MSN are static weak, then can use poisson process
for them,
params_msn_d1 - dictionary with timing and burst freq setup for msn
{'base_rates':0.1,
'base_times':[1],
'mod_rates': 20,
'mod_times':[1,200],
'mod_units':list()
'n_tot':500,
n_mod=20}
params_msn_d2 - dictionary with timing and burst freq setup for gpe
params_stn - dictionary {'rate':50}
same as params_msn
neuron_model - string, the neuron model to use
synapse_models - dict, {'MSN':'...', 'GPE':,'...', 'STN':'...'}
sim_time - simulation time
seed - seed for random generator
I_e_add - diabled
start_rec - start recording from
model_params - general model paramters
'''
params = {
'conns': {
'MSN_D1_SNR': {
'syn': synapse_models[0]
},
'GPE_SNR': {
'syn': synapse_models[1]
}
}
}
my_nest.ResetKernel(threads=8)
numpy.random.seed(seed)
params = misc.dict_merge(model_params, params)
params = misc.dict_merge({'neurons': {'GPE': {'paused': 0}}}, params)
model_list, model_dict = models({}, p_weights)
layer_list, connect_list = network(model_dict, params, p_conn)
dic_p_I_e = {'SNR': 1., 'GPE': 1., 'STN': 1.}
if p_I_e is not False:
dic_p_I_e['SNR'] *= p_I_e[0]
dic_p_I_e['GPE'] *= p_I_e[1]
dic_p_I_e['STN'] *= p_I_e[2]
# Create neurons and synapses
layer_dic = {}
for name, model, props in layer_list:
# Update input current
my_nest.MyLoadModels(model_dict, [model[1]])
if name in I_IN_VIVO.keys():
I_in_vitro = my_nest.GetDefaults(model[1])['I_e']
I_e = I_in_vitro + I_IN_VIVO[name]
my_nest.SetDefaults(model[1], {'I_e': I_e * dic_p_I_e[name]})
#! Create layer, retrieve neurons ids per elements and p
if model[0] == 'spike_generator':
layer = MyLayerPoissonInput(layer_props=props,
sd=True,
sd_params={
'start': start_rec,
'stop': sim_time
})
elif model[0] == 'poisson_generator':
layer = MyPoissonInput(model[0],
props['columns'],
sd=True,
sd_params={
'start': start_rec,
'stop': sim_time
})
else:
layer = MyLayerGroup(layer_props=props,
sd=True,
mm=False,
mm_dt=0.1,
sd_params={
'start': start_rec,
'stop': sim_time
})
for iter, id in enumerate(layer[:]):
if name == 'GPE' and params_msn_d2[
'n_mod'] and iter < params['neurons']['GPE']['paused']:
scg = my_nest.Create('step_current_generator', n=1)
rec = my_nest.GetStatus([id])[0]['receptor_types']
my_nest.SetStatus(
scg, {
'amplitude_times': params_msn_d2['mod_times'],
'amplitude_values': [0., -300., 0.]
})
my_nest.Connect(scg, [id],
params={'receptor_type': rec['CURR']})
I_e = my_nest.GetDefaults(model[1])['I_e']
if I_E_VARIATION[name]:
I = numpy.random.normal(
I_e, I_E_VARIATION[name]) #I_E_VARIATION[name])
else:
I = I_e
my_nest.SetStatus([id], {'I_e': I})
layer_dic[name] = layer
# Connect populations
for conn in connect_list:
print[conn[2]['synapse_model']]
if not conn[2]['synapse_model'] in nest.Models():
my_nest.MyLoadModels(model_dict, [conn[2]['synapse_model']])
if layer_dic[conn[0]].model == 'poisson_generator':
my_nest.Connect(layer_dic[conn[0]].ids,
layer_dic[conn[1]].ids,
model=conn[2]['synapse_model'])
else:
name = conn[0] + '_' + conn[1] + '_' + conn[3]
tp.ConnectLayers(layer_dic[conn[0]].layer_id,
layer_dic[conn[1]].layer_id, conn[2])
layer_dic[conn[1]].add_connection(source=layer_dic[conn[0]],
type=conn[3],
props=conn[2])
# Sort MSN D2 such that the closest to center is first in ids list.
# Do this to we can get focused inhibition in GPe
if params_msn_d2['focus']:
MSN_D2_idx = layer_dic['MSN_D2'].sort_ids()
else:
MSN_D2_idx = range(len(numpy.array(layer_dic['MSN_D2'].ids)))
n_mod_msn_d1 = params_msn_d1['n_mod']
n_mod_msn_d2 = params_msn_d2['n_mod']
MSN_D1_ids = layer_dic['MSN_D1'].ids
MSN_D2_ids = layer_dic['MSN_D2'].ids
MSN_D1_mod, MSN_D2_mod = [], []
if params_msn_d1['n_mod']: MSN_D1_mod = MSN_D1_ids[0:n_mod_msn_d1]
if params_msn_d2['n_mod']:
MSN_D2_mod = MSN_D2_ids[0:n_mod_msn_d2 *
params_msn_d2['skip']:params_msn_d2['skip']]
MSN_D1_base = list(set(MSN_D1_ids).difference(MSN_D1_mod))
MSN_D2_base = list(set(MSN_D2_ids).difference(MSN_D2_mod))
layer_dic['MSN_D1'].set_spike_times(params_msn_d1['base_rates'],
params_msn_d1['base_times'],
sim_time,
ids=MSN_D1_base)
layer_dic['MSN_D2'].set_spike_times(params_msn_d2['base_rates'],
params_msn_d2['base_times'],
sim_time,
ids=MSN_D2_base)
if params_msn_d1['n_mod']:
layer_dic['MSN_D1'].set_spike_times(params_msn_d1['mod_rates'],
params_msn_d1['mod_times'],
sim_time)
if params_msn_d2['n_mod']:
layer_dic['MSN_D2'].set_spike_times(params_msn_d2['mod_rates'],
params_msn_d2['mod_times'],
sim_time,
ids=MSN_D2_mod)
# If background poisson are use
if params_msn_d1['bg_rate']:
layer_dic['MSN_D1_bg'].set_spike_times(params_msn_d1['bg_rate'], [1.],
sim_time)
if params_msn_d2['bg_rate']:
layer_dic['MSN_D2_bg'].set_spike_times(params_msn_d2['bg_rate'], [1.],
sim_time)
STN_CTX_input_base = my_nest.Create('poisson_generator',
params={
'rate': BASE_RATE_CTX_STN,
'start': 0.,
'stop': sim_time
})
my_nest.MyLoadModels(model_dict, ['CTX_STN_ampa_s'])
if 'STN' in layer_dic.keys():
my_nest.DivergentConnect(STN_CTX_input_base,
layer_dic['STN'].ids,
model='CTX_STN_ampa_s')
if params_stn['mod'] and 'STN' in layer_dic.keys():
STN_CTX_input_mod = my_nest.Create('poisson_generator',
params={
'rate': params_stn['mod_rate'],
'start':
params_stn['mod_times'][0],
'stop':
params_stn['mod_times'][1]
})
my_nest.DivergentConnect(STN_CTX_input_mod,
layer_dic['STN'].ids,
model='CTX_STN_ampa_s')
my_nest.MySimulate(sim_time)
if params_msn_d1['n_mod']: layer_dic['MSN_D1'].id_mod = MSN_D1_mod
if params_msn_d2['n_mod']: layer_dic['MSN_D2'].id_mod = MSN_D2_mod
if 'MSN_D1' in layer_dic.keys():
layer_dic['MSN_D1'].get_signal('s', start=start_rec, stop=sim_time)
if 'MSN_D2' in layer_dic.keys():
layer_dic['MSN_D2'].get_signal('s', start=start_rec, stop=sim_time)
if 'GPE' in layer_dic.keys():
layer_dic['GPE'].get_signal('s', start=start_rec, stop=sim_time)
if 'SNR' in layer_dic.keys():
layer_dic['SNR'].get_signal('s', start=start_rec, stop=sim_time)
if 'STN' in layer_dic.keys():
layer_dic['STN'].get_signal('s', start=start_rec, stop=sim_time)
return layer_dic
def simulate_network_test(params_msn_d1,
params_msn_d2,
params_stn,
synapse_models,
sim_time,
seed,
I_e_add,
threads=1,
start_rec=0,
model_params={},
params_in={},
dis_conn_GPE_STN=False):
'''
params_msn_d1 - dictionary with timing and burst freq setup for msn
{'base_rates':0.1,
'base_times':[1],
'mod_rates': 20,
'mod_times':[1,200],
'mod_units':list()
'n_tot':500,
n_mod=20}
params_msn_d2 - dictionary with timing and burst freq setup for gpe
params_stn - dictionary {'rate':50}
same as params_msn
neuron_model - string, the neuron model to use
synapse_models - dict, {'MSN':'...', 'GPE':,'...', 'STN':'...'}
sim_time - simulation time
seed - seed for random generator
I_e_add - diabled
start_rec - start recording from
model_params - general model paramters
'''
my_nest.ResetKernel(threads=8)
numpy.random.seed(seed)
params = {
'conns': {
'MSN_D1_SNR': {
'syn': synapse_models[0]
},
'GPE_SNR': {
'syn': synapse_models[1]
}
}
}
params = misc.dict_merge(model_params, params)
params = misc.dict_merge({'neurons': {'GPE': {'paused': 0}}}, params)
model_list, model_dict = models(params_in)
layer_list, connect_list = network(model_dict, params)
# Create neurons and synapses
layer_dic = {}
for name, model, props in layer_list:
# Update input current
my_nest.MyLoadModels(model_dict, [model[1]])
if name in I_IN_VIVO.keys():
I_e = my_nest.GetDefaults(model[1])['I_e'] + I_IN_VIVO[name]
my_nest.SetDefaults(model[1], {'I_e': I_e})
#! Create layer, retrieve neurons ids per elements and p
if model[0] == 'spike_generator':
layer = MyLayerPoissonInput(layer_props=props,
sd=True,
sd_params={
'start': start_rec,
'stop': sim_time
})
else:
layer = MyLayerGroup(layer_props=props,
sd=True,
mm=False,
mm_dt=0.1,
sd_params={
'start': start_rec,
'stop': sim_time
})
for iter, id in enumerate(layer[:]):
if name == 'GPE' and params_msn_d2[
'n_mod'] and iter < params['neurons']['GPE']['paused']:
scg = my_nest.Create('step_current_generator', n=1)
rec = my_nest.GetStatus([id])[0]['receptor_types']
my_nest.SetStatus(
scg, {
'amplitude_times': params_msn_d2['mod_times'],
'amplitude_values': [0., -300., 0.]
})
my_nest.Connect(scg, [id],
params={'receptor_type': rec['CURR']})
I_e = my_nest.GetDefaults(model[1])['I_e']
if I_E_VARIATION[name]:
I = numpy.random.normal(I_e, I_E_VARIATION[name])
else:
I = I_e
#I=I_e
my_nest.SetStatus([id], {'I_e': I})
layer_dic[name] = layer
mm = nest.Create('multimeter', 1)
recodables = ['V_m', 'I', 'g_AMPA', 'g_NMDA', 'g_GABAA_1', 'g_GABAA_2']
my_nest.SetStatus(mm, {'interval': 0.1, 'record_from': recodables})
my_nest.Connect(mm, [layer_dic['STN'].ids[0]])
# Connect populations
for conn in connect_list:
name = conn[0] + '_' + conn[1]
my_nest.MyLoadModels(model_dict, [conn[2]['synapse_model']])
if dis_conn_GPE_STN == 'GPE' and (name in ['GPE_SNR']):
r, syn = 32 * 30.0, 'GPE_SNR_gaba_s_ref'
if not syn in my_nest.Models():
my_nest.MyLoadModels(model_dict, [syn])
pg = my_nest.Create('poisson_generator', 1, {
'rate': r,
'start': 1.
})
my_nest.DivergentConnect(pg, layer_dic[conn[1]].ids, model=syn)
elif dis_conn_GPE_STN == 'STN' and (name in ['STN_SNR']):
r, syn = 30 * 10.0, 'STN_SNR_ampa_s'
if not syn in my_nest.Models():
my_nest.MyLoadModels(model_dict, [syn])
pg = my_nest.Create('poisson_generator', 1, {
'rate': r,
'start': 1.
})
my_nest.DivergentConnect(pg, layer_dic[conn[1]].ids, model=syn)
else:
name = name + '_' + conn[3]
tp.ConnectLayers(layer_dic[conn[0]].layer_id,
layer_dic[conn[1]].layer_id, conn[2])
layer_dic[conn[1]].add_connection(source=layer_dic[conn[0]],
type=conn[3],
props=conn[2])
# Sort MSN D2 such that the closest to center is first in ids list.
# Do this to we can get focused inhibition in GPe
if params_msn_d2['focus']:
MSN_D2_idx = layer_dic['MSN_D2'].sort_ids()
else:
MSN_D2_idx = range(len(numpy.array(layer_dic['MSN_D2'].ids)))
n_mod_msn_d1 = params_msn_d1['n_mod']
n_mod_msn_d2 = params_msn_d2['n_mod']
MSN_D1_ids = layer_dic['MSN_D1'].ids
MSN_D2_ids = layer_dic['MSN_D2'].ids
MSN_D1_mod, MSN_D2_mod = [], []
if params_msn_d1['n_mod']: MSN_D1_mod = MSN_D1_ids[0:n_mod_msn_d1]
if params_msn_d2['n_mod']:
MSN_D2_mod = MSN_D2_ids[0:n_mod_msn_d2 *
params_msn_d2['skip']:params_msn_d2['skip']]
MSN_D1_base = list(set(MSN_D1_ids).difference(MSN_D1_mod))
MSN_D2_base = list(set(MSN_D2_ids).difference(MSN_D2_mod))
#layer_dic['MSN_D1'].ids[0:n_base_msn_d1]
#MSN_D2_ids=numpy.array(layer_dic['MSN_D2'].ids)
#MSN_D2_base=MSN_D2_ids#[MSN_D2_idx[0:n_base_msn_d1]]
#set().difference(t)
layer_dic['MSN_D1'].set_spike_times(params_msn_d1['base_rates'],
params_msn_d1['base_times'],
sim_time,
ids=MSN_D1_base)
layer_dic['MSN_D2'].set_spike_times(params_msn_d2['base_rates'],
params_msn_d2['base_times'],
sim_time,
ids=MSN_D2_base)
if params_msn_d1['n_mod']:
layer_dic['MSN_D1'].set_spike_times(params_msn_d1['mod_rates'],
params_msn_d1['mod_times'],
sim_time,
ids=MSN_D1_mod)
if params_msn_d2['n_mod']:
layer_dic['MSN_D2'].set_spike_times(params_msn_d2['mod_rates'],
params_msn_d2['mod_times'],
sim_time,
ids=MSN_D2_mod)
STN_CTX_input_base = my_nest.Create('poisson_generator',
params={
'rate': BASE_RATE_CTX_STN,
'start': 0.,
'stop': sim_time
})
my_nest.MyLoadModels(model_dict, ['CTX_STN_ampa_s'])
my_nest.DivergentConnect(STN_CTX_input_base,
layer_dic['STN'].ids,
model='CTX_STN_ampa_s')
if params_stn['mod']:
STN_CTX_input_mod = my_nest.Create('poisson_generator',
params={
'rate': params_stn['mod_rate'],
'start':
params_stn['mod_times'][0],
'stop':
params_stn['mod_times'][1]
})
my_nest.DivergentConnect(STN_CTX_input_mod,
layer_dic['STN'].ids,
model='CTX_STN_ampa_s')
#tar=[]
#for id in layer_dic['MSN_D1'].ids:
# tar.extend(sorted(nest.GetStatus(my_nest.FindConnections([id]),'target'))[:-1])
#pylab.subplot(211).hist(tar, 1500)
#
# tar=[]
# for id in layer_dic['MSN_D2'].ids:
# tar.extend(sorted(nest.GetStatus(my_nest.FindConnections([id]),'target'))[1:])
#
# pylab.subplot(212).hist(tar, 1500)
#pylab.show()
#
#
my_nest.MySimulate(sim_time)
if params_msn_d1['n_mod']: layer_dic['MSN_D1'].id_mod = MSN_D1_mod
if params_msn_d2['n_mod']: layer_dic['MSN_D2'].id_mod = MSN_D2_mod
#layer_dic['MSN_D1'].get_signal( 's', start=start_rec, stop=sim_time )
#layer_dic['MSN_D2'].get_signal( 's', start=start_rec, stop=sim_time )
#layer_dic['GPE'].get_signal( 's', start=start_rec, stop=sim_time )
#layer_dic['SNR'].get_signal( 's', start=start_rec, stop=sim_time )
#layer_dic['STN'].get_signal( 's', start=start_rec, stop=sim_time )
st_mm = my_nest.GetStatus(mm)[0]
pylab.plot(st_mm['events']['g_AMPA'])
pylab.plot(st_mm['events']['g_GABAA_1'])
pylab.plot(st_mm['events']['g_NMDA'])
pylab.plot(st_mm['events']['g_GABAA_2'])
m_ampa = numpy.mean(st_mm['events']['g_AMPA'])
m_gaba = numpy.mean(st_mm['events']['g_GABAA_1'])
pylab.title("{0} m_ampa:{1:2.1f} m_gaba:{2:2.1f}".format(
my_nest.version(), m_ampa, m_gaba))
pylab.show()
return layer_dic
def simulate_example_msn_snr():
nFun=0 # Function number
nSim=0 # Simulation number within function
rates=numpy.array([.1,.1])
times=numpy.array([0.,25000.])
nMSN =500
simTime=100000.
I_e=0.
my_nest.ResetKernel()
model_list=models()
my_nest.MyLoadModels( model_list, neuronModels )
my_nest.MyLoadModels( model_list, synapseModels )
MSN = MyGroup( 'spike_generator', nMSN, mm_dt=1.0, mm=False, sd=False,
spath=spath,
sname_nb=str(nFun)+str(nSim))
SNR = MyGroup( neuronModels[0], n=len(synapseModels), params={'I_e':I_e},
sd=True,
mm_dt = .1, mm=False, spath=spath,
sname_nb=str(nFun)+str(nSim) )
nSim+=1
spikeTimes=[]
for i in range(nMSN):
spikes=misc.inh_poisson_spikes( rates, times,
t_stop=simTime,
n_rep=1, seed=i )
my_nest.SetStatus([MSN[i]], params={ 'spike_times':spikes } )
for spk in spikes: spikeTimes.append((i,spk))
# add spike list for MSN to MSN spike list
MSN.signals['spikes'] = my_signals.MySpikeList(spikeTimes, MSN.ids)
MSN.save_signal( 's')
noise=my_nest.Create('noise_generator', params={'std':100.})
my_nest.Connect(noise,[SNR[0]],params={'receptor_type':5})
my_nest.Connect(noise,[SNR[1]],params={'receptor_type':5})
my_nest.Connect(noise,[SNR[2]],params={'receptor_type':5})
for i, syn in enumerate(synapseModels):
my_nest.ConvergentConnect(MSN[:],[SNR[i]], model=syn)
my_nest.MySimulate( simTime )
SNR.get_signal( 's' ) # retrieve signal
SNR_rates=[SNR.signals['spikes'].mean_rates(0,5000),
SNR.signals['spikes'].mean_rates(5000, 10000)]
for i in range(0, len(SNR_rates)):
for j in range(0, len(SNR_rates[0])):
SNR_rates[i][j]=int(SNR_rates[i][j])
s='\n'
s =s + 'Example plot MSN and SNr:\n'
s =s + 'Synapse models:\n'
for syn in synapseModels:
s = s + ' %s\n' % (syn )
s = s + ' %s %5s %3s \n' % ( 'N MSN:', str ( nMSN ), '#' )
s = s + ' %s %5s %3s \n' % ( 'MSN Rates:', str ( [str(round(r,1))
for r in rates]),'Hz' )
s = s + ' %s %5s %3s \n' % ( '\nSNR Rates 0-5000:\n',
str ( SNR_rates [0]) ,'Hz' )
s = s + ' %s %5s %3s \n' % ( '\nSNR Rates 10000-5000:\n',
str ( SNR_rates [1]) ,'Hz' )
s = s + ' %s %5s %3s \n' % ( '\nTimes:', str ( times), 'ms' )
s = s + ' %s %5s %3s \n' % ( 'I_e:', str ( I_e ), 'pA' )
infoString=s
return MSN, SNR, infoString
def simulate_GPE_vs_SNR_rate():
nFun = 1 # Function number
nSim = 0 # Simulation number within function
GPEmeanRates = numpy.arange(1, 50, 2)
SNRmeanRates = []
nGPE = 10
simTime = 10000.
I_e = 0.
for r in GPEmeanRates:
my_nest.ResetKernel()
model_list = models()
my_nest.MyLoadModels(model_list, neuronModels)
my_nest.MyLoadModels(model_list, synapseModels)
GPE = MyGroup('spike_generator',
nGPE,
mm_dt=1.0,
mm=False,
sd=False,
spath=spath,
siter=str(nFun) + str(nSim))
SNR = MyGroup(neuronModels[0],
n=len(synapseModels),
params={'I_e': I_e},
mm_dt=.1,
mm=False,
spath=spath,
siter=str(nFun) + str(nSim))
nSim += 1
if not LOAD:
spikeTimes = []
for i in range(nGPE):
spikes = misc.inh_poisson_spikes(numpy.array([r]),
numpy.array([0]),
t_stop=simTime,
n_rep=1,
seed=i)
my_nest.SetStatus([GPE[i]], params={'spike_times': spikes})
for spk in spikes:
spikeTimes.append((i, spk))
# add spike list for GPE to GPE spike list
GPE.signals['spikes'] = my_signals.MySpikeList(spikeTimes, GPE.ids)
GPE.save_signal('s')
for i, syn in enumerate(synapseModels):
my_nest.ConvergentConnect(GPE[:], [SNR[i]], model=syn)
my_nest.MySimulate(simTime)
SNR.save_signal('s')
SNR.get_signal('s') # retrieve signal
elif LOAD:
SNR.load_signal('s')
SNRmeanRates.append(SNR.signals['spikes'].mean_rates(0, simTime))
SNRmeanRates = numpy.array(SNRmeanRates).transpose()
GPEmeanRates = numpy.array(GPEmeanRates)
THR = 2.
rateAtThr = ''
for SNRr in SNRmeanRates:
tmp = str(GPEmeanRates[SNRr >= THR][-1])
rateAtThr += ' ' + tmp[0:4]
s = '\n'
s = s + 'GPE vs SNr rate:\n'
s = s + ' %s %5s %3s \n' % ('N GPEs:', str(nGPE), '#')
s = s + ' \n%s \n%5s %3s \n' % ('GPE rates:', str(GPEmeanRates[0]) + '-' +
str(GPEmeanRates[-1]), 'Hz')
s = s + ' %s %5s %3s \n' % ('Threshold SNr:', str(THR), 'Hz')
s = s + ' \n%s \n%5s %3s \n' % ('GPE rate at threshold SNr:',
str(rateAtThr), 'Hz')
s = s + ' \n%s %5s %3s \n' % ('Simulation time:', str(simTime), 'ms')
s = s + ' %s %5s %3s \n' % ('I_e:', str(I_e), 'pA')
infoString = s
return GPEmeanRates, SNRmeanRates, infoString
def simulate_example_GPE_snr():
nFun = 0 # Function number
nSim = 0 # Simulation number within function
rates = numpy.array([20, 30])
times = numpy.array([0., 5000.])
nGPE = 10
simTime = 10000.
I_e = 0.
my_nest.ResetKernel()
model_list = models()
my_nest.MyLoadModels(model_list, neuronModels)
my_nest.MyLoadModels(model_list, synapseModels)
GPE = MyGroup('spike_generator',
nGPE,
mm_dt=1.0,
mm=False,
sd=False,
spath=spath,
siter=str(nFun) + str(nSim))
SNR = MyGroup(neuronModels[0],
n=len(synapseModels),
params={'I_e': I_e},
mm_dt=.1,
mm=False,
spath=spath,
siter=str(nFun) + str(nSim))
nSim += 1
if not LOAD:
spikeTimes = []
for i in range(nGPE):
spikes = misc.inh_poisson_spikes(rates,
times,
t_stop=simTime,
n_rep=1,
seed=i)
my_nest.SetStatus([GPE[i]], params={'spike_times': spikes})
for spk in spikes:
spikeTimes.append((i, spk))
# add spike list for GPE to GPE spike list
GPE.signals['spikes'] = my_signals.MySpikeList(spikeTimes, GPE.ids)
GPE.save_signal('s')
for i, syn in enumerate(synapseModels):
my_nest.ConvergentConnect(GPE[:], [SNR[i]], model=syn)
my_nest.MySimulate(simTime)
SNR.save_signal('s')
SNR.get_signal('s') # retrieve signal
elif LOAD:
GPE.load_signal('s')
SNR.load_signal('s')
SNR_rates = [
SNR.signals['spikes'].mean_rates(0, 5000),
SNR.signals['spikes'].mean_rates(5000, 10000)
]
for i in range(0, len(SNR_rates)):
for j in range(0, len(SNR_rates[0])):
SNR_rates[i][j] = int(SNR_rates[i][j])
s = '\n'
s = s + 'Example plot GPE and SNr:\n'
s = s + 'Synapse models:\n'
for syn in synapseModels:
s = s + ' %s\n' % (syn)
s = s + ' %s %5s %3s \n' % ('N GPE:', str(nGPE), '#')
s = s + ' %s %5s %3s \n' % ('GPE Rates:',
str([str(round(r, 1)) for r in rates]), 'Hz')
s = s + ' %s %5s %3s \n' % ('\nSNR Rates 0-5000:\n', str(
SNR_rates[0]), 'Hz')
s = s + ' %s %5s %3s \n' % ('\nSNR Rates 10000-5000:\n', str(
SNR_rates[1]), 'Hz')
s = s + ' %s %5s %3s \n' % ('\nTimes:', str(times), 'ms')
s = s + ' %s %5s %3s \n' % ('I_e:', str(I_e), 'pA')
infoString = s
return GPE, SNR, infoString
def simulate_example_irregular_firing(I_vec=[0]):
simTime = 2000. # ms
my_nest.ResetKernel()
model_list, model_dict=models()
my_nest.MyLoadModels( model_list, NEURON_MODELS )
n=len(I_vec)
GPE = MyGroup( NEURON_MODELS[0], n, sd=True, mm=True, mm_dt = 1.0 )
I_e=I_vec[0]
my_nest.SetStatus(GPE[:], params={'I_e':I_e}) # Set I_e
scg = my_nest.Create( 'step_current_generator',n=n )
noise=my_nest.Create('noise_generator', params={'mean':0.,'std':10.})
rec=my_nest.GetStatus(GPE[:])[0]['receptor_types']
for source, target, I in zip(scg, GPE[:], I_vec):
#I=5.
my_nest.SetStatus([source], {'amplitude_times':[1., simTime],
'amplitude_values':[-5.,float(I)]})
my_nest.Connect( [source], [target],
params = { 'receptor_type' : rec['CURR'] } )
my_nest.Connect( noise, [target],
params = { 'receptor_type' : rec['CURR'] } )
my_nest.MySimulate(simTime)
GPE.get_signal( 'v','V_m', stop=simTime ) # retrieve signal
GPE.get_signal( 's') # retrieve signal
GPE.signals['V_m'].my_set_spike_peak( 15, spkSignal= GPE.signals['spikes'] )
#a=GPE.signals['V_m'].analog_signals[1].signal
#pylab.plot(a)
# #a=a[500:]
# pylab.subplot(211).plot(a, 'r')
#pylab.show()
#
# a=a
# a=a-numpy.mean(a)
#
# numpy.savetxt("foo.csv", a, delimiter=",")
#
# ff=numpy.abs(numpy.fft.fft(a))
# #pylab.plot(ff)
# c=numpy.correlate(a, a, mode='full')
# pylab.subplot(212).plot(c)
# pylab.show()
meanRate=round(GPE.signals['spikes'].mean_rate(0,500),1)
s='\n'
s =s + 'Example inhibitory current:\n'
s = s + ' %s %5s %3s %s %5s %3s \n' % ( 'Mean rate:', meanRate, 'Hz',
'I_e', I_e,'pA' )
s = s + 'Steps:\n'
s = s + ' %5s %3s \n' % ( I_vec, 'pA' )
infoString=s
return GPE, infoString
0.1 / 0.61] # This shoud give 1.0 Hz in std for each
# of them
f = [0.122, 0.108, 0.19]
mrs = [13.96 * f[0], 15.33 * f[1], 9.63 * f[2]]
save_result_at = OUTPUT_PATH + '/simulate.plk'
if 0:
neuron_list = []
for i, model in enumerate(neuron_models):
my_nest.MyLoadModels(model_dict, [model])
I_in_vitro = my_nest.GetDefaults(model)['I_e']
neuron = MyGroup(model, n=n, sd=True, mm_dt=.1, mm=False)
for id in neuron.ids:
I = numpy.random.normal(I_in_vitro,
I_in_vitro * norm_std[i] * mrs[i])
my_nest.SetStatus([id], {'I_e': I})
neuron_list.append(neuron)
noise = my_nest.Create('noise_generator',
params={
'mean': 0.,
'std': 1.
})
rec = my_nest.GetStatus(neuron[:])[0]['receptor_types']
for id in neuron.ids:
my_nest.Connect(noise, [id], params={'receptor_type': rec['CURR']})
my_nest.MySimulate(sim_time)
mr_list = []
for neuron in neuron_list:
