def generate(reference = "SimpleNet",
scale=1,
format='xml'):
population_size = scale_pop_size(3,scale)
nml_doc, network = oc.generate_network(reference)
oc.include_opencortex_cell(nml_doc, 'izhikevich/RS.cell.nml')
pop = oc.add_population_in_rectangular_region(network,
'RS_pop',
'RS',
population_size,
0,0,0,
100,100,100,
color='0 .8 0')
import neuroml
pop.properties.append(neuroml.Property('radius',10))
syn = oc.add_exp_two_syn(nml_doc,
id="syn0",
gbase="2nS",
erev="0mV",
tau_rise="0.5ms",
tau_decay="10ms")
pfs = oc.add_poisson_firing_synapse(nml_doc,
id="poissonFiringSyn",
average_rate="50 Hz",
synapse_id=syn.id)
oc.add_inputs_to_population(network,
"Stim0",
pop,
pfs.id,
all_cells=True)
nml_file_name = '%s.net.nml'%network.id
oc.save_network(nml_doc,
nml_file_name,
validate=(format=='xml'),
format = format)
if format=='xml':
oc.generate_lems_simulation(nml_doc,
network,
nml_file_name,
duration = 500,
dt = 0.025,
report_file_name='report.simple.txt')
def generate(cell_id, duration, reference, iRange):
cell_file = '%s.cell.nml'%cell_id
#cell_id = 'Cell0'
#cell_file = 'L23_morph.cell.nml'
nml_doc, network = oc.generate_network(reference, temperature='35degC')
oc.include_neuroml2_cell_and_channels(nml_doc,cell_file,cell_id)
population_size = len(iRange)
pop = oc.add_population_in_rectangular_region(network,
'L23_pop',
cell_id,
population_size,
0,0,0,
1000,100,1000)
for i in range(iRange.size):
stim_id = ("Stim_%i"%i)
pg = oc.add_pulse_generator(nml_doc,
id=stim_id,
delay="50ms",
duration="250ms",
amplitude="%fnA"%iRange[i])
oc.add_inputs_to_population(network,
stim_id,
pop,
pg.id,
all_cells=False,
only_cells=[i])
nml_file_name = '%s.net.nml'%network.id
oc.save_network(nml_doc, nml_file_name, validate=True)
oc.generate_lems_simulation(nml_doc,
network,
nml_file_name,
duration,
dt = 0.025)
def generate(reference="Balanced",
scalePops=1,
scalex=1,
scaley=1,
scalez=1,
connections=True,
connections_scaling=1,
duration=1000,
global_delay=2,
max_in_pop_to_plot_and_save=5,
gen_spike_saves_for_all_somas=True,
deterministic=True,
format='xml'):
num_exc = scale_pop_size(80, scalePops)
num_inh = scale_pop_size(40, scalePops)
if scalePops != 1:
reference += '_%s' % scalePops
nml_doc, network = oc.generate_network(reference)
oc.include_opencortex_cell(
nml_doc, 'AllenInstituteCellTypesDB_HH/HH_477127614.cell.nml')
oc.include_opencortex_cell(
nml_doc, 'AllenInstituteCellTypesDB_HH/HH_476686112.cell.nml')
xDim = 400 * scalex
yDim = 500 * scaley
zDim = 300 * scalez
xs = -200
ys = -150
zs = 100
##### Synapses
synAmpa1 = oc.add_exp_two_syn(nml_doc,
id="synAmpa1",
gbase="1nS",
erev="0mV",
tau_rise="0.5ms",
tau_decay="5ms")
synGaba1 = oc.add_exp_two_syn(nml_doc,
id="synGaba1",
gbase="2nS",
erev="-80mV",
tau_rise="1ms",
tau_decay="20ms")
##### Input types
if not deterministic:
pfs1 = oc.add_poisson_firing_synapse(nml_doc,
id="psf1",
average_rate="150 Hz",
synapse_id=synAmpa1.id)
##### Populations
popExc = oc.add_population_in_rectangular_region(network, 'popExc',
'HH_477127614', num_exc,
xs, ys, zs, xDim, yDim,
zDim)
popInh = oc.add_population_in_rectangular_region(network, 'popInh',
'HH_476686112', num_inh,
xs, ys, zs, xDim, yDim,
zDim)
##### Projections
total_conns = 0
if connections:
proj = oc.add_probabilistic_projection(network,
"proj0",
popExc,
popExc,
synAmpa1.id,
connections_scaling * 0.3,
delay=global_delay)
total_conns += len(proj.connection_wds)
proj = oc.add_probabilistic_projection(network,
"proj1",
popExc,
popInh,
synAmpa1.id,
connections_scaling * 0.5,
delay=global_delay)
total_conns += len(proj.connection_wds)
proj = oc.add_probabilistic_projection(network,
"proj3",
popInh,
popExc,
synGaba1.id,
connections_scaling * 0.7,
delay=global_delay)
total_conns += len(proj.connection_wds)
proj = oc.add_probabilistic_projection(network,
"proj4",
popInh,
popInh,
synGaba1.id,
connections_scaling * 0.5,
delay=global_delay)
total_conns += len(proj.connection_wds)
##### Inputs
if not deterministic:
oc.add_inputs_to_population(network,
"Stim0",
popExc,
pfs1.id,
all_cells=True)
else:
for i in range(num_exc):
pg = oc.add_pulse_generator(nml_doc,
id="pg_%i" % i,
delay="0ms",
duration="10000ms",
amplitude="%snA" % (random() * 0.5))
oc.add_inputs_to_population(network,
"Stim_%i" % i,
popExc,
pg.id,
all_cells=False,
only_cells=[i])
##### Save NeuroML and LEMS Simulation files
nml_file_name = '%s%s.net.%s' % ('XH_' if format == 'xml_hdf5' else '',
network.id,
'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format == 'xml'),
format=format)
plot_v = {popExc.id: [], popInh.id: []}
save_v = {'%s_v.dat' % popExc.id: [], '%s_v.dat' % popInh.id: []}
for i in range(min(max_in_pop_to_plot_and_save, num_exc)):
plot_v[popExc.id].append("%s/%i/%s/v" %
(popExc.id, i, popExc.component))
save_v['%s_v.dat' % popExc.id].append("%s/%i/%s/v" %
(popExc.id, i, popExc.component))
for i in range(min(max_in_pop_to_plot_and_save, num_inh)):
plot_v[popInh.id].append("%s/%i/%s/v" %
(popInh.id, i, popInh.component))
save_v['%s_v.dat' % popInh.id].append("%s/%i/%s/v" %
(popInh.id, i, popInh.component))
lems_file_name = "LEMS_%s.xml" % network.id
if format != 'xml':
lems_file_name = "LEMS_%s_%s.xml" % (network.id, format)
lems_file_name = oc.generate_lems_simulation(
nml_doc,
network,
nml_file_name,
duration=duration,
dt=0.025,
gen_plots_for_all_v=False,
gen_plots_for_quantities=plot_v,
gen_saves_for_all_v=False,
gen_saves_for_quantities=save_v,
gen_spike_saves_for_all_somas=gen_spike_saves_for_all_somas,
lems_file_name=lems_file_name)
return nml_doc, nml_file_name, lems_file_name
def generate(scalePops = 1,
percentage_exc_detailed=0,
exc2_cell = 'SmithEtAl2013/L23_Retuned_477127614',
percentage_inh_detailed=0,
scalex=1,
scaley=1,
scalez=1,
ratio_inh_exc=2,
connections=True,
exc_target_dendrites=False,
inh_target_dendrites=False,
duration = 1000,
dt = 0.025,
input_rate = 150,
global_delay = 2,
max_in_pop_to_plot_and_save = 5,
format='xml',
suffix='',
run_in_simulator = None,
num_processors = 1,
target_dir='./temp/',
exc_clamp=None): # exc_clamp is work in progress...
reference = ("Multiscale__g%s__i%s%s"%(ratio_inh_exc,input_rate,suffix)).replace('.','_')
num_exc = scale_pop_size(80,scalePops)
num_exc2 = int(math.ceil(num_exc*percentage_exc_detailed/100.0))
num_exc -= num_exc2
num_inh = scale_pop_size(40,scalePops)
num_inh2 = int(math.ceil(num_inh*percentage_inh_detailed/100.0))
num_inh -= num_inh2
nml_doc, network = oc.generate_network(reference, network_seed=1234)
#exc_cell_id = 'AllenHH_480351780'
exc_cell_id = 'AllenHH_477127614'
#exc_cell_id = 'HH_477127614'
exc_type = exc_cell_id.split('_')[0]
oc.include_neuroml2_cell_and_channels(nml_doc, 'cells/%s/%s.cell.nml'%(exc_type,exc_cell_id), exc_cell_id)
#inh_cell_id = 'AllenHH_485058595'
inh_cell_id = 'AllenHH_476686112'
#inh_cell_id = 'HH_476686112'
inh_type = exc_cell_id.split('_')[0]
oc.include_neuroml2_cell_and_channels(nml_doc, 'cells/%s/%s.cell.nml'%(inh_type,inh_cell_id), inh_cell_id)
if percentage_exc_detailed>0:
exc2_cell_id = exc2_cell.split('/')[1]
exc2_cell_dir = exc2_cell.split('/')[0]
oc.include_neuroml2_cell_and_channels(nml_doc, 'cells/%s/%s.cell.nml'%(exc2_cell_dir,exc2_cell_id), exc2_cell_id)
if percentage_inh_detailed>0:
inh2_cell_id = 'cNAC187_L23_NBC_9d37c4b1f8_0_0'
oc.include_neuroml2_cell_and_channels(nml_doc, 'cells/BBP/%s.cell.nml'%inh2_cell_id, inh2_cell_id)
xDim = 700*scalex
yDim = 200*scaley
zDim = 700*scalez
xs = -1*xDim/2
ys = -1*yDim/2
zs = -1*zDim/2
##### Synapses
exc_syn_nS = 1.
synAmpa1 = oc.add_exp_two_syn(nml_doc, id="synAmpa1", gbase="%snS"%exc_syn_nS,
erev="0mV", tau_rise="0.5ms", tau_decay="5ms")
synGaba1 = oc.add_exp_two_syn(nml_doc, id="synGaba1", gbase="%snS"%(exc_syn_nS*ratio_inh_exc),
erev="-80mV", tau_rise="1ms", tau_decay="20ms")
##### Input types
pfs1 = oc.add_poisson_firing_synapse(nml_doc,
id="psf1",
average_rate="%s Hz"%input_rate,
synapse_id=synAmpa1.id)
##### Populations
popExc = oc.add_population_in_rectangular_region(network,
'popExc',
exc_cell_id,
num_exc,
xs,ys,zs,
xDim,yDim,zDim,
color=exc_color)
allExc = [popExc]
if num_exc2>0:
popExc2 = oc.add_population_in_rectangular_region(network,
'popExc2',
exc2_cell_id,
num_exc2,
xs,ys,zs,
xDim,yDim,zDim,
color=exc2_color)
allExc.append(popExc2)
popInh = oc.add_population_in_rectangular_region(network,
'popInh',
inh_cell_id,
num_inh,
xs,ys,zs,
xDim,yDim,zDim,
color=inh_color)
allInh = [popInh]
if num_inh2>0:
popInh2 = oc.add_population_in_rectangular_region(network,
'popInh2',
inh2_cell_id,
num_inh2,
xs,ys,zs,
xDim,yDim,zDim,
color=inh2_color)
allInh.append(popInh2)
##### Projections
if connections:
exc_exc_conn_prob = 0.5
exc_inh_conn_prob = 0.7
inh_exc_conn_prob = 0.7
inh_inh_conn_prob = 0.5
for popEpr in allExc:
for popEpo in allExc:
proj = add_projection(network, "projEE",
popEpr, popEpo,
synAmpa1.id, exc_exc_conn_prob,
global_delay,
exc_target_dendrites)
for popIpo in allInh:
proj = add_projection(network, "projEI",
popEpr, popIpo,
synAmpa1.id, exc_inh_conn_prob,
global_delay,
exc_target_dendrites)
for popIpr in allInh:
for popEpo in allExc:
proj = add_projection(network, "projIE",
popIpr, popEpo,
synGaba1.id, inh_exc_conn_prob,
global_delay,
inh_target_dendrites)
for popIpo in allInh:
proj = add_projection(network, "projII",
popIpr, popIpo,
synGaba1.id, inh_inh_conn_prob,
global_delay,
inh_target_dendrites)
##### Inputs
for pop in allExc:
oc.add_inputs_to_population(network, "Stim_%s"%pop.id,
pop, pfs1.id,
all_cells=True)
# Work in progress...
# General idea: clamp one (or more) exc cell at rev pot of inh syn and see only exc inputs
#
if exc_clamp:
vc = oc.add_voltage_clamp_triple(nml_doc, id="exc_clamp",
delay='0ms',
duration='%sms'%duration,
conditioning_voltage=synGaba1.erev,
testing_voltage=synGaba1.erev,
return_voltage=synGaba1.erev,
simple_series_resistance="1e5ohm",
active = "1")
for pop in exc_clamp:
oc.add_inputs_to_population(network, "exc_clamp_%s"%pop,
network.get_by_id(pop), vc.id,
all_cells=False,
only_cells=exc_clamp[pop])
##### Save NeuroML and LEMS Simulation files
nml_file_name = '%s.net.%s'%(network.id,'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format=='xml'),
format = format,
target_dir=target_dir)
if format=='xml':
save_v = {}
plot_v = {}
if num_exc>0:
exc_traces = '%s_%s_v.dat'%(network.id,popExc.id)
save_v[exc_traces] = []
plot_v[popExc.id] = []
if num_inh>0:
inh_traces = '%s_%s_v.dat'%(network.id,popInh.id)
save_v[inh_traces] = []
plot_v[popInh.id] = []
if num_exc2>0:
exc2_traces = '%s_%s_v.dat'%(network.id,popExc2.id)
save_v[exc2_traces] = []
plot_v[popExc2.id] = []
if num_inh2>0:
inh2_traces = '%s_%s_v.dat'%(network.id,popInh2.id)
save_v[inh2_traces] = []
plot_v[popInh2.id] = []
for i in range(min(max_in_pop_to_plot_and_save,num_exc)):
plot_v[popExc.id].append("%s/%i/%s/v"%(popExc.id,i,popExc.component))
save_v[exc_traces].append("%s/%i/%s/v"%(popExc.id,i,popExc.component))
for i in range(min(max_in_pop_to_plot_and_save,num_exc2)):
plot_v[popExc2.id].append("%s/%i/%s/v"%(popExc2.id,i,popExc2.component))
save_v[exc2_traces].append("%s/%i/%s/v"%(popExc2.id,i,popExc2.component))
for i in range(min(max_in_pop_to_plot_and_save,num_inh)):
plot_v[popInh.id].append("%s/%i/%s/v"%(popInh.id,i,popInh.component))
save_v[inh_traces].append("%s/%i/%s/v"%(popInh.id,i,popInh.component))
for i in range(min(max_in_pop_to_plot_and_save,num_inh2)):
plot_v[popInh2.id].append("%s/%i/%s/v"%(popInh2.id,i,popInh2.component))
save_v[inh2_traces].append("%s/%i/%s/v"%(popInh2.id,i,popInh2.component))
gen_spike_saves_for_all_somas = run_in_simulator!='jNeuroML_NetPyNE'
lems_file_name = oc.generate_lems_simulation(nml_doc, network,
target_dir+nml_file_name,
duration = duration,
dt = dt,
gen_plots_for_all_v = False,
gen_plots_for_quantities = plot_v,
gen_saves_for_all_v = False,
gen_saves_for_quantities = save_v,
gen_spike_saves_for_all_somas = gen_spike_saves_for_all_somas,
target_dir=target_dir)
if run_in_simulator:
print ("Running %s for %sms in %s"%(lems_file_name, duration, run_in_simulator))
traces, events = oc.simulate_network(lems_file_name,
run_in_simulator,
max_memory='4000M',
nogui=True,
load_saved_data=True,
reload_events=True,
plot=False,
verbose=False,
num_processors=num_processors)
print("Reloaded traces: %s"%traces.keys())
#print("Reloaded events: %s"%events.keys())
use_events_for_rates = False
exc_rate = 0
inh_rate = 0
if use_events_for_rates:
if (run_in_simulator=='jNeuroML_NetPyNE'):
raise('Saving of spikes (and so calculation of rates) not yet supported in jNeuroML_NetPyNE')
for ek in events.keys():
rate = 1000 * len(events[ek])/float(duration)
print("Cell %s has rate %s Hz"%(ek,rate))
if 'popExc' in ek:
exc_rate += rate/num_exc
if 'popInh' in ek:
inh_rate += rate/num_inh
else:
tot_exc_rate = 0
exc_cells = 0
tot_inh_rate = 0
inh_cells = 0
tt = [t*1000 for t in traces['t']]
for tk in traces.keys():
if tk!='t':
rate = get_rate_from_trace(tt,[v*1000 for v in traces[tk]])
print("Cell %s has rate %s Hz"%(tk,rate))
if 'popExc' in tk:
tot_exc_rate += rate
exc_cells+=1
if 'popInh' in tk:
tot_inh_rate += rate
inh_cells+=1
exc_rate = tot_exc_rate/exc_cells
inh_rate = tot_inh_rate/inh_cells
print("Run %s: Exc rate: %s Hz; Inh rate %s Hz"%(reference,exc_rate, inh_rate))
return exc_rate, inh_rate, traces
else:
lems_file_name = None
return nml_doc, nml_file_name, lems_file_name
proj2 = oc.add_targeted_projection(network,
prefix="proj2",
presynaptic_population=popPyr,
postsynaptic_population=popPyrS,
targeting_mode="convergent",
synapse_list=[synAmpa1.id],
number_conns_per_cell=2,
pre_segment_group="soma_group",
post_segment_group="soma_group",
delays_dict={synAmpa1.id:3},
weights_dict={synAmpa1.id:'3*random()'})
##### Inputs
oc.add_inputs_to_population(network, "Stim0",
popIzh, pfsStrong.id,
all_cells=True)
oc.add_inputs_to_population(network, "Stim1",
popIaf, pfs200.id,
all_cells=True)
oc.add_inputs_to_population(network, "Stim2",
popIafRef, pfs200.id,
all_cells=True)
oc.add_inputs_to_population(network, "Stim3",
popPyrS, pfs100.id,
all_cells=True)
def generate(reference="GapJunctions",
num_pre=5,
num_post=2,
connections=True,
duration=1000,
segments_to_plot_record={
'pop_pre': [0],
'pop_post': [0]
},
format='xml'):
nml_doc, network = oc.generate_network(reference)
oc.include_opencortex_cell(nml_doc, 'acnet2/pyr_4_sym.cell.nml')
xDim = 500
yDim = 50
zDim = 500
pop_pre = oc.add_population_in_rectangular_region(network, 'pop_pre',
'pyr_4_sym', num_pre, 0,
0, 0, xDim, yDim, zDim)
pop_post = oc.add_population_in_rectangular_region(network, 'pop_post',
'pyr_4_sym', num_post,
0, yDim, 0, xDim,
yDim + yDim, zDim)
ampa_syn = oc.add_exp_two_syn(nml_doc,
id="AMPA_syn",
gbase="30e-9S",
erev="0mV",
tau_rise="0.003s",
tau_decay="0.0031s")
gj_syn = oc.add_gap_junction_synapse(nml_doc, id="gj0", conductance="5nS")
pfs = oc.add_poisson_firing_synapse(nml_doc,
id="poissonFiringSyn",
average_rate="30 Hz",
synapse_id=ampa_syn.id)
oc.add_inputs_to_population(network,
"Stim0",
pop_pre,
pfs.id,
all_cells=True)
total_conns = 0
if connections:
this_syn = gj_syn.id
proj = oc.add_targeted_electrical_projection(
nml_doc,
network,
"Proj0",
pop_pre,
pop_post,
targeting_mode='convergent',
synapse_list=[this_syn],
pre_segment_group='soma_group',
post_segment_group='dendrite_group',
number_conns_per_cell=3)
if proj:
total_conns += len(proj[0].electrical_connections) + len(
proj[0].electrical_connection_instances)
nml_file_name = '%s.net.%s' % (network.id,
'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format == 'xml'),
format=format)
if format == 'xml':
gen_plots_for_quantities = {
} # Dict with displays vs lists of quantity paths
gen_saves_for_quantities = {
} # Dict with file names vs lists of quantity paths
for pop in segments_to_plot_record.keys():
pop_nml = network.get_by_id(pop)
if pop_nml is not None and pop_nml.size > 0:
group = len(segments_to_plot_record[pop]) == 1
if group:
display = 'Display_%s_v' % (pop)
file_ = 'Sim_%s.%s.v.dat' % (nml_doc.id, pop)
gen_plots_for_quantities[display] = []
gen_saves_for_quantities[file_] = []
for i in range(int(pop_nml.size)):
if not group:
display = 'Display_%s_%i_v' % (pop, i)
file_ = 'Sim_%s.%s.%i.v.dat' % (nml_doc.id, pop, i)
gen_plots_for_quantities[display] = []
gen_saves_for_quantities[file_] = []
for seg in segments_to_plot_record[pop]:
quantity = '%s/%i/%s/%i/v' % (pop, i,
pop_nml.component, seg)
gen_plots_for_quantities[display].append(quantity)
gen_saves_for_quantities[file_].append(quantity)
lems_file_name = oc.generate_lems_simulation(
nml_doc,
network,
nml_file_name,
duration=duration,
dt=0.025,
gen_plots_for_all_v=False,
gen_plots_for_quantities=gen_plots_for_quantities,
gen_saves_for_all_v=False,
gen_saves_for_quantities=gen_saves_for_quantities)
else:
lems_file_name = None
return nml_doc, nml_file_name, lems_file_name
syn1 = oc.add_exp_two_syn(nml_doc,
id="syn1",
gbase="2nS",
erev="0mV",
tau_rise="1ms",
tau_decay="15ms")
pfs = oc.add_poisson_firing_synapse(nml_doc,
id="poissonFiringSyn",
average_rate="150 Hz",
synapse_id=syn0.id)
oc.add_inputs_to_population(network,
"Stim0",
pop_pre,
pfs.id,
all_cells=True)
oc.add_probabilistic_projection(network,
"proj0",
pop_pre,
pop_post,
syn1.id,
0.3,
weight=0.05,
delay=5)
duration = 1000
dt = 0.01
nml_file_name = '%s.net.nml'%network.id
def generate(scale_populations=1,
percentage_exc_detailed=0,
exc2_cell='SmithEtAl2013/L23_Retuned_477127614',
percentage_inh_detailed=0,
scalex=1,
scaley=1,
scalez=1,
Bee=.1,
Bei=.1,
Bie=-.2,
Bii=-.2,
Be_bkg=.1,
Be_stim=.1,
r_bkg=0,
r_stim=0,
fraction_inh_pert=0.75,
connections=True,
exc_target_dendrites=False,
inh_target_dendrites=False,
duration=1000,
dt=0.025,
global_delay=.1,
max_in_pop_to_plot_and_save=10,
format='xml',
suffix='',
run_in_simulator=None,
num_processors=1,
target_dir='./temp/',
exc_clamp=None): # exc_clamp is work in progress...
reference = ("ISN_net%s" % (suffix)).replace('.', '_')
print('-------------------------------------------------')
print(' Generating ISN network: %s' % reference)
print(' Duration: %s; dt: %s; scale: %s; simulator: %s (num proc. %s)' %
(duration, dt, scale_populations, run_in_simulator, num_processors))
print(' Bee: %s; Bei: %s; Bie: %s; Bii: %s' % (Bee, Bei, Bie, Bii))
print(' Be_bkg: %s at %sHz' % (Be_bkg, r_bkg))
print(' Be_stim: %s at %sHz (i.e. %sHz for %s perturbed I cells)' %
(Be_stim, r_stim, r_bkg + r_stim, fraction_inh_pert))
print(' Exc detailed: %s%% - Inh detailed %s%%' %
(percentage_exc_detailed, percentage_inh_detailed))
print('-------------------------------------------------')
num_exc = scale_pop_size(80, scale_populations)
num_exc2 = int(math.ceil(num_exc * percentage_exc_detailed / 100.0))
num_exc -= num_exc2
num_inh = scale_pop_size(20, scale_populations)
num_inh2 = int(math.ceil(num_inh * percentage_inh_detailed / 100.0))
num_inh -= num_inh2
nml_doc, network = oc.generate_network(reference, network_seed=1234)
#exc_cell_id = 'AllenHH_480351780'
exc_cell_id = 'AllenHH_477127614'
#exc_cell_id = 'HH_477127614'
exc_cell_id = 'HH2_477127614'
exc_type = exc_cell_id.split('_')[0]
oc.include_neuroml2_cell_and_channels(
nml_doc, 'cells/%s/%s.cell.nml' % (exc_type, exc_cell_id), exc_cell_id)
#inh_cell_id = 'AllenHH_485058595'
inh_cell_id = 'AllenHH_476686112'
#inh_cell_id = 'AllenHH_477127614'
#inh_cell_id = 'HH_476686112'
inh_cell_id = 'HH2_476686112'
inh_type = exc_cell_id.split('_')[0]
oc.include_neuroml2_cell_and_channels(
nml_doc, 'cells/%s/%s.cell.nml' % (inh_type, inh_cell_id), inh_cell_id)
if percentage_exc_detailed > 0:
exc2_cell_id = exc2_cell.split('/')[1]
exc2_cell_dir = exc2_cell.split('/')[0]
oc.include_neuroml2_cell_and_channels(
nml_doc, 'cells/%s/%s.cell.nml' % (exc2_cell_dir, exc2_cell_id),
exc2_cell_id)
if percentage_inh_detailed > 0:
inh2_cell_id = 'cNAC187_L23_NBC_9d37c4b1f8_0_0'
oc.include_neuroml2_cell_and_channels(
nml_doc, 'cells/BBP/%s.cell.nml' % inh2_cell_id, inh2_cell_id)
xDim = 700 * scalex
yDim = 200 * scaley
zDim = 700 * scalez
xs = -1 * xDim / 2
ys = -1 * yDim / 2
zs = -1 * zDim / 2
##### Synapses
synAmpaEE = oc.add_exp_one_syn(nml_doc,
id="synAmpaEE",
gbase="%snS" % Bee,
erev="0mV",
tau_decay="1ms")
synAmpaEI = oc.add_exp_one_syn(nml_doc,
id="synAmpaEI",
gbase="%snS" % Bei,
erev="0mV",
tau_decay="1ms")
synGabaIE = oc.add_exp_one_syn(nml_doc,
id="synGabaIE",
gbase="%snS" % abs(Bie),
erev="-80mV",
tau_decay="1ms")
synGabaII = oc.add_exp_one_syn(nml_doc,
id="synGabaII",
gbase="%snS" % abs(Bii),
erev="-80mV",
tau_decay="1ms")
synAmpaBkg = oc.add_exp_one_syn(nml_doc,
id="synAmpaBkg",
gbase="%snS" % Be_bkg,
erev="0mV",
tau_decay="1ms")
synAmpaStim = oc.add_exp_one_syn(nml_doc,
id="synAmpaStim",
gbase="%snS" % Be_stim,
erev="0mV",
tau_decay="1ms")
##### Input types
tpfsA = oc.add_transient_poisson_firing_synapse(
nml_doc,
id="tpsfA",
average_rate="%s Hz" % r_bkg,
delay='0ms',
duration='%sms' % (Ttrans + Tblank),
synapse_id=synAmpaBkg.id)
tpfsB = oc.add_transient_poisson_firing_synapse(
nml_doc,
id="tpsfB",
average_rate="%s Hz" % r_bkg,
delay='%sms' % (Ttrans + Tblank),
duration='%sms' % (Tstim),
synapse_id=synAmpaBkg.id)
tpfsC = oc.add_transient_poisson_firing_synapse(
nml_doc,
id="tpsfC",
average_rate="%s Hz" % (r_bkg + r_stim),
delay='%sms' % (Ttrans + Tblank),
duration='%sms' % (Tstim),
synapse_id=synAmpaStim.id)
##### Populations
popExc = oc.add_population_in_rectangular_region(network,
'popExc',
exc_cell_id,
num_exc,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color=exc_color)
allExc = [popExc]
if num_exc2 > 0:
popExc2 = oc.add_population_in_rectangular_region(network,
'popExc2',
exc2_cell_id,
num_exc2,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color=exc2_color)
allExc.append(popExc2)
popInh = oc.add_population_in_rectangular_region(network,
'popInh',
inh_cell_id,
num_inh,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color=inh_color)
allInh = [popInh]
if num_inh2 > 0:
popInh2 = oc.add_population_in_rectangular_region(network,
'popInh2',
inh2_cell_id,
num_inh2,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color=inh2_color)
allInh.append(popInh2)
##### Projections
if connections:
exc_exc_conn_prob = 0.15
exc_inh_conn_prob = 0.15
inh_exc_conn_prob = 1
inh_inh_conn_prob = 1
weight_expr = 'abs(normal(1,0.5))'
for popEpr in allExc:
for popEpo in allExc:
proj = add_projection(network, "projEE", popEpr, popEpo,
synAmpaEE.id, exc_exc_conn_prob,
global_delay, exc_target_dendrites,
weight_expr)
for popIpo in allInh:
proj = add_projection(network, "projEI", popEpr, popIpo,
synAmpaEI.id, exc_inh_conn_prob,
global_delay, exc_target_dendrites,
weight_expr)
for popIpr in allInh:
for popEpo in allExc:
proj = add_projection(network, "projIE", popIpr, popEpo,
synGabaIE.id, inh_exc_conn_prob,
global_delay, inh_target_dendrites,
weight_expr)
for popIpo in allInh:
proj = add_projection(network, "projII", popIpr, popIpo,
synGabaII.id, inh_inh_conn_prob,
global_delay, inh_target_dendrites,
weight_expr)
##### Inputs
for pop in allExc:
oc.add_inputs_to_population(network,
"Stim_pre_%s" % pop.id,
pop,
tpfsA.id,
all_cells=True)
for pop in allInh:
oc.add_inputs_to_population(network,
"Stim_pre_%s" % pop.id,
pop,
tpfsA.id,
all_cells=True)
for pop in allExc:
oc.add_inputs_to_population(network,
"Stim_E_%s" % pop.id,
pop,
tpfsB.id,
all_cells=True)
for pop in allInh:
num_inh_pert = int(pop.get_size() * fraction_inh_pert)
oc.add_inputs_to_population(network,
"Stim_I_pert_%s" % pop.id,
pop,
tpfsC.id,
all_cells=False,
only_cells=range(0, num_inh_pert))
oc.add_inputs_to_population(network,
"Stim_I_nonpert_%s" % pop.id,
pop,
tpfsB.id,
all_cells=False,
only_cells=range(num_inh_pert,
pop.get_size()))
# Work in progress...
# General idea: clamp one (or more) exc cell at rev pot of inh syn and see only exc inputs
#
if exc_clamp:
vc = oc.add_voltage_clamp_triple(nml_doc,
id="exc_clamp",
delay='0ms',
duration='%sms' % duration,
conditioning_voltage=synGaba1.erev,
testing_voltage=synGaba1.erev,
return_voltage=synGaba1.erev,
simple_series_resistance="1e5ohm",
active="1")
for pop in exc_clamp:
oc.add_inputs_to_population(network,
"exc_clamp_%s" % pop,
network.get_by_id(pop),
vc.id,
all_cells=False,
only_cells=exc_clamp[pop])
##### Save NeuroML and LEMS Simulation files
nml_file_name = '%s.net.%s' % (network.id,
'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format == 'xml'),
format=format,
target_dir=target_dir)
print("Saved to: %s" % nml_file_name)
save_v = {}
plot_v = {}
if num_exc > 0:
exc_traces = '%s_%s_v.dat' % (network.id, popExc.id)
save_v[exc_traces] = []
plot_v[popExc.id] = []
if num_inh > 0:
inh_traces = '%s_%s_v.dat' % (network.id, popInh.id)
save_v[inh_traces] = []
plot_v[popInh.id] = []
if num_exc2 > 0:
exc2_traces = '%s_%s_v.dat' % (network.id, popExc2.id)
save_v[exc2_traces] = []
plot_v[popExc2.id] = []
if num_inh2 > 0:
inh2_traces = '%s_%s_v.dat' % (network.id, popInh2.id)
save_v[inh2_traces] = []
plot_v[popInh2.id] = []
for i in range(min(max_in_pop_to_plot_and_save, num_exc)):
plot_v[popExc.id].append("%s/%i/%s/v" %
(popExc.id, i, popExc.component))
save_v[exc_traces].append("%s/%i/%s/v" %
(popExc.id, i, popExc.component))
for i in range(min(max_in_pop_to_plot_and_save, num_exc2)):
plot_v[popExc2.id].append("%s/%i/%s/v" %
(popExc2.id, i, popExc2.component))
save_v[exc2_traces].append("%s/%i/%s/v" %
(popExc2.id, i, popExc2.component))
for i in range(min(max_in_pop_to_plot_and_save, num_inh)):
plot_v[popInh.id].append("%s/%i/%s/v" %
(popInh.id, i, popInh.component))
save_v[inh_traces].append("%s/%i/%s/v" %
(popInh.id, i, popInh.component))
for i in range(min(max_in_pop_to_plot_and_save, num_inh2)):
plot_v[popInh2.id].append("%s/%i/%s/v" %
(popInh2.id, i, popInh2.component))
save_v[inh2_traces].append("%s/%i/%s/v" %
(popInh2.id, i, popInh2.component))
gen_spike_saves_for_all_somas = True
lems_file_name, lems_sim = oc.generate_lems_simulation(
nml_doc,
network,
target_dir + nml_file_name,
duration=duration,
dt=dt,
gen_plots_for_all_v=False,
gen_plots_for_quantities=plot_v,
gen_saves_for_all_v=False,
gen_saves_for_quantities=save_v,
gen_spike_saves_for_all_somas=gen_spike_saves_for_all_somas,
target_dir=target_dir,
report_file_name='report.txt')
if run_in_simulator:
print("Running %s for %sms in %s" %
(lems_file_name, duration, run_in_simulator))
traces, events = oc.simulate_network(lems_file_name,
run_in_simulator,
max_memory='4000M',
nogui=True,
load_saved_data=True,
reload_events=True,
plot=False,
verbose=True,
num_processors=num_processors)
print("Reloaded traces: %s" % traces.keys())
#print("Reloaded events: %s"%events.keys())
use_events_for_rates = False
exc_rate = 0
inh_rate = 0
if use_events_for_rates:
if (run_in_simulator == 'jNeuroML_NetPyNE'):
raise (
'Saving of spikes (and so calculation of rates) not yet supported in jNeuroML_NetPyNE'
)
for ek in events.keys():
rate = 1000 * len(events[ek]) / float(duration)
print("Cell %s has a rate %s Hz" % (ek, rate))
if 'popExc' in ek:
exc_rate += rate / num_exc
if 'popInh' in ek:
inh_rate += rate / num_inh
else:
tot_exc_rate = 0
exc_cells = 0
tot_inh_rate = 0
inh_cells = 0
tt = [t * 1000 for t in traces['t']]
for tk in traces.keys():
if tk != 't':
rate = get_rate_from_trace(tt,
[v * 1000 for v in traces[tk]])
print("Cell %s has rate %s Hz" % (tk, rate))
if 'popExc' in tk:
tot_exc_rate += rate
exc_cells += 1
if 'popInh' in tk:
tot_inh_rate += rate
inh_cells += 1
exc_rate = tot_exc_rate / exc_cells
inh_rate = tot_inh_rate / inh_cells
print("Run %s: Exc rate: %s Hz; Inh rate %s Hz" %
(reference, exc_rate, inh_rate))
return exc_rate, inh_rate, traces
return nml_doc, nml_file_name, lems_file_name
zDim)
offset += yDim
pop_rs = oc.add_population_in_rectangular_region(network, 'pop_rs', 'RS', 5, 0,
offset, 0, xDim, yDim, zDim)
##### Projections
oc.add_probabilistic_projection(network, "proj0", pop_iaf, pop_rs, synAmpa1.id,
0.5)
##### Inputs
oc.add_inputs_to_population(network,
"Stim0",
pop_iaf,
pg0.id,
only_cells=[i for i in [0, 2] if i < pop_iaf.size])
oc.add_inputs_to_population(network,
"Stim1",
pop_iaf,
pg1.id,
only_cells=[i for i in [3, 4] if i < pop_iaf.size])
nml_file_name = '%s.net.nml' % network.id
oc.save_network(nml_doc, nml_file_name, validate=True)
oc.generate_lems_simulation(nml_doc,
network,
nml_file_name,
def generate(reference="L23TraubDemo",
num_rs=DEFAULT_RS_POP_SIZE,
num_bask=DEFAULT_BASK_POP_SIZE,
scalex=1,
scaley=1,
scalez=1,
connections=False,
poisson_inputs=True,
offset_curent_range_pA=None,
global_delay=0,
duration=300,
segments_to_plot_record={
'pop_rs': [0],
'pop_bask': [0]
},
format='xml'):
nml_doc, network = oc.generate_network(reference)
#oc.add_cell_and_channels(nml_doc, 'acnet2/pyr_4_sym.cell.nml','pyr_4_sym')
oc.include_opencortex_cell(nml_doc, 'Thalamocortical/L23PyrRS.cell.nml')
oc.include_opencortex_cell(nml_doc, 'Thalamocortical/SupBasket.cell.nml')
xDim = 500 * scalex
yDim = 200 * scaley
zDim = 500 * scalez
pop_rs = oc.add_population_in_rectangular_region(network, 'pop_rs',
'L23PyrRS', num_rs, 0, 0,
0, xDim, yDim, zDim)
pop_bask = oc.add_population_in_rectangular_region(network, 'pop_bask',
'SupBasket', num_bask,
0, 0, 0, xDim, yDim,
zDim)
syn0 = oc.add_exp_two_syn(nml_doc,
id="syn0",
gbase="1nS",
erev="0mV",
tau_rise="0.5ms",
tau_decay="10ms")
syn1 = oc.add_exp_two_syn(nml_doc,
id="syn1",
gbase="2nS",
erev="0mV",
tau_rise="1ms",
tau_decay="15ms")
if poisson_inputs:
pfs = oc.add_poisson_firing_synapse(nml_doc,
id="poissonFiringSyn",
average_rate="150 Hz",
synapse_id=syn0.id)
oc.add_inputs_to_population(network,
"Stim0",
pop_rs,
pfs.id,
all_cells=True)
'''
oc.add_inputs_to_population(network,
"Stim1",
pop_bask,
pfs.id,
all_cells=True)'''
if offset_curent_range_pA:
for pop_id in offset_curent_range_pA:
pop = next(p for p in network.populations if p.id == pop_id)
pg0 = oc.add_pulse_generator(nml_doc,
id="offset_current_%s" % pop.id,
delay="0ms",
duration="%sms" % duration,
amplitude="1pA")
import neuroml
import random
input_list = neuroml.InputList(id="inputs_offset_current_%s" %
pop.id,
component=pg0.id,
populations=pop.id)
network.input_lists.append(input_list)
min_, max_ = offset_curent_range_pA[pop_id]
for i in range(pop.get_size()):
input = neuroml.InputW(
id=i,
target="../%s/%i/%s" % (pop.id, i, pop.component),
destination="synapses",
weight=(min_ + (max_ - min_) * random.random()))
input_list.input_ws.append(input)
total_conns = 0
if connections:
proj = oc.add_probabilistic_projection(network,
"proj0",
pop_rs,
pop_bask,
syn1.id,
1,
weight=1,
delay=global_delay)
'''
proj = oc.add_targeted_projection(nml_doc,
network,
"proj0",
presynaptic_population = pop_rs,
postsynaptic_population = pop_bask,
targeting_mode = 'convergent',
synapse_list = [syn1.id],
number_conns_per_cell = 1,
pre_segment_group = 'axon_group',
post_segment_group = 'dendrite_group',
delays_dict = {syn1.id:2},
weights_dict = {syn1.id:1})'''
if proj:
total_conns += len(proj.connection_wds)
if num_rs != DEFAULT_RS_POP_SIZE or num_bask != DEFAULT_BASK_POP_SIZE:
new_reference = '%s_%scells_%sconns' % (nml_doc.id, num_rs + num_bask,
total_conns)
network.id = new_reference
nml_doc.id = new_reference
nml_file_name = '%s.net.%s' % (network.id,
'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format == 'xml'),
format=format)
if format == 'xml':
gen_plots_for_quantities = {
} # Dict with displays vs lists of quantity paths
gen_saves_for_quantities = {
} # Dict with file names vs lists of quantity paths
for pop in segments_to_plot_record.keys():
pop_nml = network.get_by_id(pop)
if pop_nml is not None and pop_nml.size > 0:
for i in range(int(pop_nml.size)):
gen_plots_for_quantities['Display_%s_%i_v' % (pop, i)] = []
gen_saves_for_quantities['Sim_%s.%s.%i.v.dat' %
(nml_doc.id, pop, i)] = []
for seg in segments_to_plot_record[pop]:
quantity = '%s/%i/%s/%i/v' % (pop, i,
pop_nml.component, seg)
gen_plots_for_quantities['Display_%s_%i_v' %
(pop, i)].append(quantity)
gen_saves_for_quantities['Sim_%s.%s.%i.v.dat' %
(nml_doc.id, pop,
i)].append(quantity)
lems_file_name = oc.generate_lems_simulation(
nml_doc,
network,
nml_file_name,
duration=duration,
dt=0.025,
gen_plots_for_all_v=False,
gen_plots_for_quantities=gen_plots_for_quantities,
gen_saves_for_all_v=False,
gen_saves_for_quantities=gen_saves_for_quantities)
else:
lems_file_name = None
return nml_doc, nml_file_name, lems_file_name
def generate(reference="VClamp",
poisson_inputs=True,
use_vclamp=False,
duration=500,
format='xml'):
nml_doc, network = oc.generate_network(reference)
oc.include_opencortex_cell(nml_doc, 'Thalamocortical/L23PyrRS.cell.nml')
num_cells = 4
pop_rs = oc.add_population_in_rectangular_region(network,
'popRS',
'L23PyrRS',
num_cells,
0,
0,
0,
1000,
20,
20,
color='.8 0 0')
syn0 = oc.add_exp_two_syn(nml_doc,
id="syn0",
gbase="1nS",
erev="0mV",
tau_rise="0.5ms",
tau_decay="10ms")
if poisson_inputs:
pfs = oc.add_transient_poisson_firing_synapse(nml_doc,
id="poissonFiringSyn",
average_rate="20 Hz",
delay="50 ms",
duration="400 ms",
synapse_id=syn0.id)
oc.add_targeted_inputs_to_population(network,
"pfs_noise",
pop_rs,
pfs.id,
segment_group='dendrite_group',
number_per_cell=100,
all_cells=True)
all_vclamp_segs = [0, 142, 87]
vclamp_segs = {0: [], 1: [0], 2: [0, 142], 3: all_vclamp_segs}
gen_plots_for_quantities = {
} # Dict with displays vs lists of quantity paths
gen_saves_for_quantities = {
} # Dict with file names vs lists of quantity paths
if use_vclamp:
v_clamped = '-70mV'
for cell_id in vclamp_segs:
for seg_id in vclamp_segs[cell_id]:
vc = oc.add_voltage_clamp_triple(
nml_doc,
id='vclamp_cell%i_seg%i' % (cell_id, seg_id),
delay='0ms',
duration='%sms' % duration,
conditioning_voltage=v_clamped,
testing_voltage=v_clamped,
return_voltage=v_clamped,
simple_series_resistance="1e1ohm",
active="1")
vc_dat_file = 'v_clamps_i_cell%s_seg%s.dat' % (cell_id, seg_id)
gen_saves_for_quantities[vc_dat_file] = []
oc.add_inputs_to_population(network,
"input_vClamp_cell%i_seg%i" %
(cell_id, seg_id),
pop_rs,
vc.id,
all_cells=False,
only_cells=[cell_id],
segment_ids=[seg_id])
q = '%s/%s/%s/%s/%s/i' % (pop_rs.id, cell_id, pop_rs.component,
seg_id, vc.id)
gen_saves_for_quantities[vc_dat_file].append(q)
nml_file_name = '%s.net.%s' % (network.id,
'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format == 'xml'),
format=format)
segments_to_plot_record = {pop_rs.id: all_vclamp_segs + [20, 50, 99, 139]}
if format == 'xml':
for pop in segments_to_plot_record.keys():
pop_nml = network.get_by_id(pop)
if pop_nml is not None and pop_nml.size > 0:
for i in range(int(pop_nml.size)):
gen_plots_for_quantities['Display_%s_%i_v' % (pop, i)] = []
gen_saves_for_quantities['Sim_%s.%s.%i.v.dat' %
(nml_doc.id, pop, i)] = []
for seg in segments_to_plot_record[pop]:
quantity = '%s/%i/%s/%i/v' % (pop, i,
pop_nml.component, seg)
gen_plots_for_quantities['Display_%s_%i_v' %
(pop, i)].append(quantity)
gen_saves_for_quantities['Sim_%s.%s.%i.v.dat' %
(nml_doc.id, pop,
i)].append(quantity)
lems_file_name = oc.generate_lems_simulation(
nml_doc,
network,
nml_file_name,
duration=duration,
dt=0.025,
gen_plots_for_all_v=False,
gen_plots_for_quantities=gen_plots_for_quantities,
gen_saves_for_all_v=False,
gen_saves_for_quantities=gen_saves_for_quantities)
else:
lems_file_name = None
return nml_doc, nml_file_name, lems_file_name
def generate(reference = "Weights",
num_each = 6,
connections=True,
duration = 1000,
format='xml'):
nml_doc, network = oc.generate_network(reference)
cell_id = 'HH_477127614'
cell = oc.include_opencortex_cell(nml_doc, 'AllenInstituteCellTypesDB_HH/%s.cell.nml'%cell_id)
xDim = 500
yDim = 500
zDim = 30
pop_pre = oc.add_population_in_rectangular_region(network, 'pop_pre',
cell_id, num_each,
0,0,0, xDim,yDim,zDim,
color='.8 0 0')
pop_post_chem_exc = oc.add_population_in_rectangular_region(network, 'pop_post_chem_exc',
cell_id, num_each+1,
0,yDim,0, xDim,yDim,zDim,
color='0 0 .8')
pop_post_chem_inh = oc.add_population_in_rectangular_region(network, 'pop_post_chem_inh',
cell_id, num_each+2,
xDim,yDim,0, xDim,yDim,zDim,
color='0 .8 .8')
pop_post_cont = oc.add_population_in_rectangular_region(network, 'pop_post_cont',
cell_id, num_each+3,
xDim,0,0, xDim,yDim,zDim,
color='0 .8 0')
ampa_syn = oc.add_exp_two_syn(nml_doc, id="AMPA_syn",
gbase="10nS", erev="0mV",
tau_rise="2ms", tau_decay="10ms")
gaba_syn = oc.add_exp_two_syn(nml_doc, id="GABA_syn",
gbase="10nS", erev="-80mV",
tau_rise="3ms", tau_decay="30ms")
gj_syn = oc.add_gap_junction_synapse(nml_doc, id="gj0",
conductance=".05nS")
analog_syn = GradedSynapse(id='analog_syn',
conductance="10nS",
delta="5mV",
Vth="-35mV",
k="0.025per_ms",
erev="0mV")
silent_syn = SilentSynapse(id="silent1")
nml_doc.graded_synapses.append(analog_syn)
nml_doc.silent_synapses.append(silent_syn)
pfs = oc.add_poisson_firing_synapse(nml_doc, id="poissonFiringSyn",
average_rate="10 Hz", synapse_id=ampa_syn.id)
oc.add_inputs_to_population(network, "Stim0",
pop_pre, pfs.id, all_cells=True)
if connections:
proj_chem_exc = oc.add_probabilistic_projection(network,
"proj_chem_exc",
pop_pre,
pop_post_chem_exc,
ampa_syn.id,
0.7,
weight=1,
delay=5)
for conn in proj_chem_exc.connection_wds:
if conn.get_pre_cell_id() < 3 and conn.get_post_cell_id() < 3:
conn.weight = 0.5
proj_chem_inh = oc.add_probabilistic_projection(network,
"proj_chem_inh",
pop_pre,
pop_post_chem_inh,
gaba_syn.id,
0.7,
weight=1,
delay=5)
for conn in proj_chem_inh.connection_wds:
if conn.get_pre_cell_id() < 3 and conn.get_post_cell_id() < 3:
conn.weight = 2
proj_cont = ContinuousProjection(id='proj_cont', \
presynaptic_population=pop_pre.id,
postsynaptic_population=pop_post_cont.id)
network.continuous_projections.append(proj_cont)
for i in range(pop_pre.get_size()):
for j in range(pop_post_cont.get_size()):
conn0 = ContinuousConnectionInstanceW(id='%s'%(j+i*pop_pre.get_size()), \
pre_cell='../%s/%s/%s'%(pop_pre.id,i,cell_id),
post_cell='../%s/%s/%s'%(pop_post_cont.id,j,cell_id),
pre_component=silent_syn.id,
post_component=analog_syn.id,
weight=(i+j)/10.0)
proj_cont.continuous_connection_instance_ws.append(conn0)
gj_pops = [pop_pre, pop_post_chem_exc, pop_post_chem_inh, pop_post_cont]
for pre in gj_pops:
for post in gj_pops:
proj_gap = oc.add_targeted_electrical_projection(nml_doc,
network,
"proj_",
pre,
post,
targeting_mode='convergent',
synapse_list=[gj_syn.id],
pre_segment_group = 'soma_group',
post_segment_group = 'soma_group',
number_conns_per_cell=3)
for conn in network.electrical_projections[-1].electrical_connection_instance_ws:
conn.weight = conn.get_pre_cell_id() + conn.get_post_cell_id()
nml_file_name = '%s.net.%s'%(network.id,'nml.h5' if format == 'hdf5' else 'nml')
target_dir = 'HDF5/' if format == 'hdf5' else './'
oc.save_network(nml_doc,
nml_file_name,
validate=(format=='xml'),
format = format,
target_dir=target_dir)
if format=='xml':
lems_file_name = oc.generate_lems_simulation(nml_doc, network,
nml_file_name,
duration = duration,
dt = 0.025,
gen_plots_for_all_v = True,
gen_saves_for_all_v = True)
else:
lems_file_name = None
return nml_doc, nml_file_name, lems_file_name
offset,
0,
xDim,
yDim,
zDim,
color='0 .8 0')
offset += yDim
##### Projections
oc.add_probabilistic_projection(network, "proj0", pop0, pop0, synAmpa1.id, 0.5)
##### Inputs
oc.add_inputs_to_population(network, "Stim0", pop0, pg0.id, only_cells=[0])
oc.add_inputs_to_population(network, "Stim1", pop0, pg1.id, only_cells=[1])
oc.add_inputs_to_population(network, "Stim2", pop0, pfs.id, all_cells=True)
nml_file_name = '%s.net.nml' % network.id
oc.save_network(nml_doc, nml_file_name, validate=True)
oc.generate_lems_simulation(nml_doc,
network,
nml_file_name,
duration=500,
dt=0.005,
plot_all_segments=True,
save_all_segments=True)
def generate(reference="Balanced",
num_bbp=1,
scalePops=1,
scalex=1,
scaley=1,
scalez=1,
connections=True,
duration=1000,
input_rate=150,
global_delay=0,
max_in_pop_to_plot_and_save=5,
gen_spike_saves_for_all_somas=True,
format='xml'):
num_exc = scale_pop_size(80, scalePops)
num_inh = scale_pop_size(40, scalePops)
nml_doc, network = oc.generate_network(reference)
oc.include_opencortex_cell(
nml_doc, 'AllenInstituteCellTypesDB_HH/HH_477127614.cell.nml')
oc.include_opencortex_cell(
nml_doc, 'AllenInstituteCellTypesDB_HH/HH_476686112.cell.nml')
if num_bbp > 0:
oc.include_opencortex_cell(
nml_doc,
'BlueBrainProject_NMC/cADpyr229_L23_PC_5ecbf9b163_0_0.cell.nml')
xDim = 400 * scalex
yDim = 500 * scaley
zDim = 300 * scalez
xs = -200
ys = -150
zs = 100
##### Synapses
synAmpa1 = oc.add_exp_two_syn(nml_doc,
id="synAmpa1",
gbase="1nS",
erev="0mV",
tau_rise="0.5ms",
tau_decay="5ms")
synGaba1 = oc.add_exp_two_syn(nml_doc,
id="synGaba1",
gbase="2nS",
erev="-80mV",
tau_rise="1ms",
tau_decay="20ms")
##### Input types
pfs1 = oc.add_poisson_firing_synapse(nml_doc,
id="psf1",
average_rate="%s Hz" % input_rate,
synapse_id=synAmpa1.id)
##### Populations
popExc = oc.add_population_in_rectangular_region(network,
'popExc',
'HH_477127614',
num_exc,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color='.8 0 0')
popInh = oc.add_population_in_rectangular_region(network,
'popInh',
'HH_476686112',
num_inh,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color='0 0 .8')
if num_bbp == 1:
popBBP = oc.add_single_cell_population(
network,
'popBBP',
'cADpyr229_L23_PC_5ecbf9b163_0_0',
z=200,
color='0 .8 0')
elif num_bbp > 1:
popBBP = oc.add_population_in_rectangular_region(
network,
'popBBP',
'cADpyr229_L23_PC_5ecbf9b163_0_0',
num_bbp,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color='0 .8 0')
##### Projections
total_conns = 0
if connections:
proj = oc.add_probabilistic_projection(network,
"proj0",
popExc,
popExc,
synAmpa1.id,
0.5,
delay=global_delay)
total_conns += len(proj.connection_wds)
proj = oc.add_probabilistic_projection(network,
"proj1",
popExc,
popInh,
synAmpa1.id,
0.7,
delay=global_delay)
total_conns += len(proj.connection_wds)
proj = oc.add_probabilistic_projection(network,
"proj3",
popInh,
popExc,
synGaba1.id,
0.7,
delay=global_delay)
total_conns += len(proj.connection_wds)
proj = oc.add_probabilistic_projection(network,
"proj4",
popInh,
popInh,
synGaba1.id,
0.5,
delay=global_delay)
total_conns += len(proj.connection_wds)
if num_bbp > 0:
proj = oc.add_probabilistic_projection(network,
"proj5",
popExc,
popBBP,
synAmpa1.id,
0.5,
delay=global_delay)
total_conns += len(proj.connection_wds)
##### Inputs
oc.add_inputs_to_population(network,
"Stim0",
popExc,
pfs1.id,
all_cells=True)
##### Save NeuroML and LEMS Simulation files
if num_bbp != 1:
new_reference = 'Balanced_%scells_%sconns' % (num_bbp + num_exc +
num_inh, total_conns)
network.id = new_reference
nml_doc.id = new_reference
nml_file_name = '%s.net.%s' % (network.id,
'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format == 'xml'),
format=format)
if format == 'xml':
plot_v = {popExc.id: [], popInh.id: []}
save_v = {'%s_v.dat' % popExc.id: [], '%s_v.dat' % popInh.id: []}
if num_bbp > 0:
plot_v[popBBP.id] = []
save_v['%s_v.dat' % popBBP.id] = []
for i in range(min(max_in_pop_to_plot_and_save, num_exc)):
plot_v[popExc.id].append("%s/%i/%s/v" %
(popExc.id, i, popExc.component))
save_v['%s_v.dat' % popExc.id].append(
"%s/%i/%s/v" % (popExc.id, i, popExc.component))
for i in range(min(max_in_pop_to_plot_and_save, num_inh)):
plot_v[popInh.id].append("%s/%i/%s/v" %
(popInh.id, i, popInh.component))
save_v['%s_v.dat' % popInh.id].append(
"%s/%i/%s/v" % (popInh.id, i, popInh.component))
for i in range(min(max_in_pop_to_plot_and_save, num_bbp)):
plot_v[popBBP.id].append("%s/%i/%s/v" %
(popBBP.id, i, popBBP.component))
save_v['%s_v.dat' % popBBP.id].append(
"%s/%i/%s/v" % (popBBP.id, i, popBBP.component))
lems_file_name = oc.generate_lems_simulation(
nml_doc,
network,
nml_file_name,
duration=duration,
dt=0.025,
gen_plots_for_all_v=False,
gen_plots_for_quantities=plot_v,
gen_saves_for_all_v=False,
gen_saves_for_quantities=save_v,
gen_spike_saves_for_all_somas=gen_spike_saves_for_all_somas)
else:
lems_file_name = None
return nml_doc, nml_file_name, lems_file_name
##### Projections
oc.add_probabilistic_projection(network,
"proj1",
pop0,
pop1,
'AMPA_NMDA',
0.7)
##### Inputs
oc.add_inputs_to_population(network,
"Stim0",
pop0,
pfs100.id,
all_cells=True,
number_per_cell=5)
oc.add_inputs_to_population(network,
"Stim1",
pop0,
pg0.id,
all_cells=True)
##### Save NeuroML and LEMS Simulation files
nml_file_name = '%s.net.nml'%network.id
oc.save_network(nml_doc, nml_file_name, validate=True)
duration="300ms",
amplitude="0.5nA")
pgHH = oc.add_pulse_generator(nml_doc,
id="pgHH",
delay="100ms",
duration="300ms",
amplitude="0.7nA")
'''
pgBBP = oc.add_pulse_generator(nml_doc,
id="pgBBP",
delay="100ms",
duration="300ms",
amplitude="0.7nA")'''
oc.add_inputs_to_population(network, "Stim0", popIzh, pgIzh.id, all_cells=True)
oc.add_inputs_to_population(network, "Stim1", popHH, pgHH.id, all_cells=True)
'''
oc.add_inputs_to_population(network,
"Stim2",
popBBP,
pgBBP.id,
all_cells=True)'''
duration = 500
dt = 0.005
nml_file_name = '%s.net.nml' % network.id
oc.save_network(nml_doc, nml_file_name, validate=True)
def generate(scalePops=1,
percentage_exc_detailed=0,
scalex=1,
scaley=1,
scalez=1,
ratio_inh_exc=2,
connections=True,
duration=1000,
input_rate=150,
global_delay=2,
max_in_pop_to_plot_and_save=5,
format='xml',
run_in_simulator=None):
reference = ("Multiscale__g%s__i%s" % (ratio_inh_exc, input_rate)).replace(
'.', '_')
num_exc = scale_pop_size(80, scalePops)
num_exc2 = int(0.5 + num_exc * percentage_exc_detailed / 100.0)
num_exc -= num_exc2
num_inh = scale_pop_size(40, scalePops)
nml_doc, network = oc.generate_network(reference)
oc.include_opencortex_cell(
nml_doc, 'AllenInstituteCellTypesDB_HH/HH_477127614.cell.nml')
oc.include_opencortex_cell(
nml_doc, 'AllenInstituteCellTypesDB_HH/HH_476686112.cell.nml')
oc.include_opencortex_cell(nml_doc,
'L23Pyr_SmithEtAl2013/L23_NoHotSpot.cell.nml')
xDim = 1000 * scalex
yDim = 300 * scaley
zDim = 1000 * scalez
xs = -200
ys = -150
zs = 100
##### Synapses
exc_syn_nS = 1.
synAmpa1 = oc.add_exp_two_syn(nml_doc,
id="synAmpa1",
gbase="%snS" % exc_syn_nS,
erev="0mV",
tau_rise="0.5ms",
tau_decay="5ms")
synGaba1 = oc.add_exp_two_syn(nml_doc,
id="synGaba1",
gbase="%snS" % (exc_syn_nS * ratio_inh_exc),
erev="-80mV",
tau_rise="1ms",
tau_decay="20ms")
##### Input types
pfs1 = oc.add_poisson_firing_synapse(nml_doc,
id="psf1",
average_rate="%s Hz" % input_rate,
synapse_id=synAmpa1.id)
##### Populations
popExc = oc.add_population_in_rectangular_region(network,
'popExc',
'HH_477127614',
num_exc,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color='0 0 1')
popExc2 = oc.add_population_in_rectangular_region(network,
'popExc2',
'L23_NoHotSpot',
num_exc2,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color='0 1 0')
allExc = [popExc, popExc2]
popInh = oc.add_population_in_rectangular_region(network,
'popInh',
'HH_476686112',
num_inh,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color='1 0 0')
##### Projections
if connections:
for pop1 in allExc:
for pop2 in allExc:
proj = oc.add_probabilistic_projection(network,
"proj0",
pop1,
pop2,
synAmpa1.id,
0.5,
delay=global_delay)
proj = oc.add_probabilistic_projection(network,
"proj1",
pop1,
popInh,
synAmpa1.id,
0.7,
delay=global_delay)
proj = oc.add_probabilistic_projection(network,
"proj2",
popInh,
pop1,
synGaba1.id,
0.7,
delay=global_delay)
proj = oc.add_probabilistic_projection(network,
"proj3",
popInh,
popInh,
synGaba1.id,
0.5,
delay=global_delay)
##### Inputs
for pop in allExc:
oc.add_inputs_to_population(network,
"Stim_%s" % pop.id,
pop,
pfs1.id,
all_cells=True)
##### Save NeuroML and LEMS Simulation files
target_dir = './temp/'
nml_file_name = '%s.net.%s' % (network.id,
'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format == 'xml'),
format=format,
target_dir=target_dir)
if format == 'xml':
plot_v = {popExc.id: [], popExc2.id: [], popInh.id: []}
exc_traces = '%s_%s_v.dat' % (network.id, popExc.id)
exc2_traces = '%s_%s_v.dat' % (network.id, popExc2.id)
inh_traces = '%s_%s_v.dat' % (network.id, popInh.id)
save_v = {exc_traces: [], inh_traces: [], exc2_traces: []}
for i in range(min(max_in_pop_to_plot_and_save, num_exc)):
plot_v[popExc.id].append("%s/%i/%s/v" %
(popExc.id, i, popExc.component))
save_v[exc_traces].append("%s/%i/%s/v" %
(popExc.id, i, popExc.component))
for i in range(min(max_in_pop_to_plot_and_save, num_exc2)):
plot_v[popExc2.id].append("%s/%i/%s/v" %
(popExc2.id, i, popExc2.component))
save_v[exc2_traces].append("%s/%i/%s/v" %
(popExc2.id, i, popExc2.component))
for i in range(min(max_in_pop_to_plot_and_save, num_inh)):
plot_v[popInh.id].append("%s/%i/%s/v" %
(popInh.id, i, popInh.component))
save_v[inh_traces].append("%s/%i/%s/v" %
(popInh.id, i, popInh.component))
gen_spike_saves_for_all_somas = run_in_simulator != 'jNeuroML_NetPyNE'
lems_file_name = oc.generate_lems_simulation(
nml_doc,
network,
target_dir + nml_file_name,
duration=duration,
dt=0.025,
gen_plots_for_all_v=False,
gen_plots_for_quantities=plot_v,
gen_saves_for_all_v=False,
gen_saves_for_quantities=save_v,
gen_spike_saves_for_all_somas=gen_spike_saves_for_all_somas,
target_dir=target_dir)
if run_in_simulator:
print("Running %s in %s" % (lems_file_name, run_in_simulator))
traces, events = oc.simulate_network(lems_file_name,
run_in_simulator,
max_memory='4000M',
nogui=True,
load_saved_data=True,
reload_events=True,
plot=False,
verbose=False)
print("Reloaded traces: %s" % traces.keys())
#print("Reloaded events: %s"%events.keys())
use_events_for_rates = False
exc_rate = 0
inh_rate = 0
if use_events_for_rates:
if (run_in_simulator == 'jNeuroML_NetPyNE'):
raise (
'Saving of spikes (and so calculation of rates) not yet supported in jNeuroML_NetPyNE'
)
for ek in events.keys():
rate = 1000 * len(events[ek]) / float(duration)
print("Cell %s has rate %s Hz" % (ek, rate))
if 'popExc' in ek:
exc_rate += rate / num_exc
if 'popInh' in ek:
inh_rate += rate / num_inh
else:
tot_exc_rate = 0
exc_cells = 0
tot_inh_rate = 0
inh_cells = 0
tt = [t * 1000 for t in traces['t']]
for tk in traces.keys():
if tk != 't':
rate = get_rate_from_trace(
tt, [v * 1000 for v in traces[tk]])
print("Cell %s has rate %s Hz" % (tk, rate))
if 'popExc' in tk:
tot_exc_rate += rate
exc_cells += 1
if 'popInh' in tk:
tot_inh_rate += rate
inh_cells += 1
exc_rate = tot_exc_rate / exc_cells
inh_rate = tot_inh_rate / inh_cells
print("Run %s: Exc rate: %s Hz; Inh rate %s Hz" %
(reference, exc_rate, inh_rate))
return exc_rate, inh_rate, traces
else:
lems_file_name = None
return nml_doc, nml_file_name, lems_file_name
def generate(reference = "Balanced",
scalePops = 1,
scalex=1,
scaley=1,
scalez=1,
connections=True,
connections_scaling=1,
duration = 1000,
global_delay = 2,
max_in_pop_to_plot_and_save = 5,
gen_spike_saves_for_all_somas = True,
deterministic = True,
format='xml'):
num_exc = scale_pop_size(80,scalePops)
num_inh = scale_pop_size(40,scalePops)
if scalePops!=1:
reference += '_%s'%scalePops
nml_doc, network = oc.generate_network(reference)
oc.include_opencortex_cell(nml_doc, 'AllenInstituteCellTypesDB_HH/HH_477127614.cell.nml')
oc.include_opencortex_cell(nml_doc, 'AllenInstituteCellTypesDB_HH/HH_476686112.cell.nml')
xDim = 400*scalex
yDim = 500*scaley
zDim = 300*scalez
xs = -200
ys = -150
zs = 100
##### Synapses
synAmpa1 = oc.add_exp_two_syn(nml_doc, id="synAmpa1", gbase="1nS",
erev="0mV", tau_rise="0.5ms", tau_decay="5ms")
synGaba1 = oc.add_exp_two_syn(nml_doc, id="synGaba1", gbase="2nS",
erev="-80mV", tau_rise="1ms", tau_decay="20ms")
##### Input types
if not deterministic:
pfs1 = oc.add_poisson_firing_synapse(nml_doc,
id="psf1",
average_rate="150 Hz",
synapse_id=synAmpa1.id)
##### Populations
popExc = oc.add_population_in_rectangular_region(network,
'popExc',
'HH_477127614',
num_exc,
xs,ys,zs,
xDim,yDim,zDim)
popInh = oc.add_population_in_rectangular_region(network,
'popInh',
'HH_476686112',
num_inh,
xs,ys,zs,
xDim,yDim,zDim)
##### Projections
total_conns = 0
if connections:
proj = oc.add_probabilistic_projection(network, "proj0",
popExc, popExc,
synAmpa1.id, connections_scaling*0.3, delay = global_delay)
total_conns += len(proj.connection_wds)
proj = oc.add_probabilistic_projection(network, "proj1",
popExc, popInh,
synAmpa1.id, connections_scaling*0.5, delay = global_delay)
total_conns += len(proj.connection_wds)
proj = oc.add_probabilistic_projection(network, "proj3",
popInh, popExc,
synGaba1.id, connections_scaling*0.7, delay = global_delay)
total_conns += len(proj.connection_wds)
proj = oc.add_probabilistic_projection(network, "proj4",
popInh, popInh,
synGaba1.id, connections_scaling*0.5, delay = global_delay)
total_conns += len(proj.connection_wds)
##### Inputs
if not deterministic:
oc.add_inputs_to_population(network, "Stim0",
popExc, pfs1.id,
all_cells=True)
else:
for i in range(num_exc):
pg = oc.add_pulse_generator(nml_doc,
id="pg_%i"%i,
delay="0ms",
duration="10000ms",
amplitude="%snA"%(random()*0.5))
oc.add_inputs_to_population(network, "Stim_%i"%i,
popExc, pg.id,
all_cells=False,
only_cells=[i])
##### Save NeuroML and LEMS Simulation files
nml_file_name = '%s%s.net.%s'%('XH_' if format == 'xml_hdf5' else '', network.id,'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format=='xml'),
format = format)
plot_v = {popExc.id:[],popInh.id:[]}
save_v = {'%s_v.dat'%popExc.id:[],'%s_v.dat'%popInh.id:[]}
for i in range(min(max_in_pop_to_plot_and_save,num_exc)):
plot_v[popExc.id].append("%s/%i/%s/v"%(popExc.id,i,popExc.component))
save_v['%s_v.dat'%popExc.id].append("%s/%i/%s/v"%(popExc.id,i,popExc.component))
for i in range(min(max_in_pop_to_plot_and_save,num_inh)):
plot_v[popInh.id].append("%s/%i/%s/v"%(popInh.id,i,popInh.component))
save_v['%s_v.dat'%popInh.id].append("%s/%i/%s/v"%(popInh.id,i,popInh.component))
lems_file_name = "LEMS_%s.xml"%network.id
if format != 'xml':
lems_file_name = "LEMS_%s_%s.xml"%(network.id,format)
lems_file_name = oc.generate_lems_simulation(nml_doc, network,
nml_file_name,
duration = duration,
dt = 0.025,
gen_plots_for_all_v = False,
gen_plots_for_quantities = plot_v,
gen_saves_for_all_v = False,
gen_saves_for_quantities = save_v,
gen_spike_saves_for_all_somas = gen_spike_saves_for_all_somas,
lems_file_name = lems_file_name)
return nml_doc, nml_file_name, lems_file_name
def generate(
scale_populations=1,
percentage_exc_detailed=0,
#exc2_cell = 'SmithEtAl2013/L23_Retuned_477127614',
exc2_cell='SmithEtAl2013/L23_NoHotSpot',
#exc2_cell = 'BBP/cADpyr229_L23_PC_5ecbf9b163_0_0',
#exc2_cell = 'BBP/cNAC187_L23_NBC_9d37c4b1f8_0_0',
#exc2_cell = 'Thalamocortical/L23PyrRS',
percentage_inh_detailed=0,
scalex=1,
scaley=1,
scalez=1,
exc_exc_conn_prob=0.25,
exc_inh_conn_prob=0.25,
inh_exc_conn_prob=0.75,
inh_inh_conn_prob=0.75,
ee2_conn_prob=0,
ie2_conn_prob=0,
Bee=.1,
Bei=.1,
Bie=-.2,
Bii=-.2,
Bee2=1,
Bie2=-2,
Be_bkg=.1,
Be_stim=.1,
r_bkg=0,
r_bkg_ExtExc=0,
r_bkg_ExtInh=0,
r_bkg_ExtExc2=0,
r_stim=0,
fraction_inh_pert=0.75,
Ttrans=500, # transitent time to discard the data (ms)
Tblank=1500, # simulation time before perturbation (ms)
Tstim=1500, # simulation time of perturbation (ms)
Tpost=500, # simulation time after perturbation (ms)
connections=True,
connections2=False,
exc_target_dendrites=False,
inh_target_dendrites=False,
duration=1000,
dt=0.025,
global_delay=.1,
max_in_pop_to_plot_and_save=10,
format='xml',
suffix='',
run_in_simulator=None,
num_processors=1,
target_dir='./temp/',
v_clamp=False,
simulation_seed=11111):
reference = ("ISN_net%s" % (suffix)).replace('.', '_')
ks = open('kernelseed', 'w')
ks.write('%i' % simulation_seed)
ks.close()
info = (' Generating ISN network: %s\n' % reference)
info += (
' Duration: %s; dt: %s; scale: %s; simulator: %s (num proc. %s)\n' %
(duration, dt, scale_populations, run_in_simulator, num_processors))
info += (' Bee: %s; Bei: %s; Bie: %s; Bii: %s\n' % (Bee, Bei, Bie, Bii))
info += (' Bkg exc at %sHz\n' % (r_bkg_ExtExc))
info += (' Bkg inh at %sHz\n' % (r_bkg_ExtInh))
info += (' Be_stim: %s at %sHz (i.e. %sHz for %s perturbed I cells)\n' %
(Be_stim, r_stim, r_bkg_ExtInh + r_stim, fraction_inh_pert))
info += (' Exc detailed: %s%% - Inh detailed %s%%\n' %
(percentage_exc_detailed, percentage_inh_detailed))
info += (' Seed: %s' % (simulation_seed))
print('-------------------------------------------------')
print(info)
print('-------------------------------------------------')
num_exc = scale_pop_size(np.round(100 * exc_inh_fraction),
scale_populations)
num_exc2 = int(math.ceil(num_exc * percentage_exc_detailed / 100.0))
num_exc -= num_exc2
num_inh = scale_pop_size(np.round(100 * (1 - exc_inh_fraction)),
scale_populations)
num_inh2 = int(math.ceil(num_inh * percentage_inh_detailed / 100.0))
num_inh -= num_inh2
nml_doc, network = oc.generate_network(reference,
network_seed=simulation_seed)
nml_doc.notes = info
network.notes = info
#exc_cell_id = 'AllenHH_480351780'
#exc_cell_id = 'AllenHH_477127614'
#exc_cell_id = 'HH_477127614'
exc_cell_id = 'HH2_477127614'
exc_type = exc_cell_id.split('_')[0]
oc.include_neuroml2_cell_and_channels(
nml_doc, 'cells/%s/%s.cell.nml' % (exc_type, exc_cell_id), exc_cell_id)
#inh_cell_id = 'AllenHH_485058595'
#inh_cell_id = 'AllenHH_476686112'
#inh_cell_id = 'AllenHH_477127614'
#inh_cell_id = 'HH_476686112'
inh_cell_id = 'HH2_476686112'
inh_type = exc_cell_id.split('_')[0]
oc.include_neuroml2_cell_and_channels(
nml_doc, 'cells/%s/%s.cell.nml' % (inh_type, inh_cell_id), inh_cell_id)
if percentage_exc_detailed > 0:
exc2_cell_id = exc2_cell.split('/')[1]
exc2_cell_dir = exc2_cell.split('/')[0]
oc.include_neuroml2_cell_and_channels(
nml_doc, 'cells/%s/%s.cell.nml' % (exc2_cell_dir, exc2_cell_id),
exc2_cell_id)
if percentage_inh_detailed > 0:
inh2_cell_id = 'cNAC187_L23_NBC_9d37c4b1f8_0_0'
oc.include_neuroml2_cell_and_channels(
nml_doc, 'cells/BBP/%s.cell.nml' % inh2_cell_id, inh2_cell_id)
xDim = 700 * scalex
yDim = 100 * scaley
yDimExc2 = 50 * scaley
zDim = 700 * scalez
xs = -1 * xDim / 2
ys = -1 * yDim / 2
zs = -1 * zDim / 2
##### Synapses
synAmpaEE = oc.add_exp_one_syn(nml_doc,
id="ampaEE",
gbase="%snS" % Bee,
erev="0mV",
tau_decay="1ms")
synAmpaEI = oc.add_exp_one_syn(nml_doc,
id="ampaEI",
gbase="%snS" % Bei,
erev="0mV",
tau_decay="1ms")
synGabaIE = oc.add_exp_one_syn(nml_doc,
id="gabaIE",
gbase="%snS" % abs(Bie),
erev="-80mV",
tau_decay="2ms")
synGabaII = oc.add_exp_one_syn(nml_doc,
id="gabaII",
gbase="%snS" % abs(Bii),
erev="-80mV",
tau_decay="2ms")
synAmpaBkg = oc.add_exp_one_syn(nml_doc,
id="ampaBkg",
gbase="%snS" % Be_bkg,
erev="0mV",
tau_decay="1ms")
#synAmpaStim = oc.add_exp_one_syn(nml_doc, id="ampaStim", gbase="%snS"%Be_stim,
# erev="0mV", tau_decay="1ms")
synAmpaEE2 = oc.add_exp_one_syn(nml_doc,
id="ampaEE2",
gbase="%snS" % Bee2,
erev="0mV",
tau_decay="10ms")
synGabaIE2 = oc.add_exp_one_syn(nml_doc,
id="gabaIE2",
gbase="%snS" % abs(Bie2),
erev="-80mV",
tau_decay="10ms")
##### Input types
'''tpfsA = oc.add_transient_poisson_firing_synapse(nml_doc,
id="tpsfA",
average_rate="%s Hz"%r_bkg,
delay = '0ms',
duration = '%sms'%(Ttrans+Tblank),
synapse_id=synAmpaBkg.id)
tpfsB = oc.add_transient_poisson_firing_synapse(nml_doc,
id="tpsfB",
average_rate="%s Hz"%r_bkg,
delay = '%sms'%(Ttrans+Tblank),
duration = '%sms'%(Tstim),
synapse_id=synAmpaBkg.id)
tpfsC = oc.add_transient_poisson_firing_synapse(nml_doc,
id="tpsfC",
average_rate="%s Hz"%(r_bkg+r_stim),
delay = '%sms'%(Ttrans+Tblank),
duration = '%sms'%(Tstim),
synapse_id=synAmpaStim.id)'''
tpfsExtExc = oc.add_transient_poisson_firing_synapse(
nml_doc,
id="tpfsExtExc",
average_rate="%s Hz" % r_bkg_ExtExc,
delay='0ms',
duration='%sms' % (Ttrans + Tblank + Tstim + Tpost),
synapse_id=synAmpaBkg.id)
tpfsExtExc2 = oc.add_transient_poisson_firing_synapse(
nml_doc,
id="tpfsExtExc2",
average_rate="%s Hz" % r_bkg_ExtExc2,
delay='0ms',
duration='%sms' % (Ttrans + Tblank + Tstim + Tpost),
synapse_id=synAmpaBkg.id)
tpfsExtInh = oc.add_transient_poisson_firing_synapse(
nml_doc,
id="tpfsExtInh",
average_rate="%s Hz" % r_bkg_ExtInh,
delay='0ms',
duration='%sms' % (Ttrans + Tblank + Tstim + Tpost),
synapse_id=synAmpaBkg.id)
tpfsPertInh_before = oc.add_transient_poisson_firing_synapse(
nml_doc,
id="tpfsPertInh_before",
average_rate="%s Hz" % r_bkg_ExtInh,
delay='0ms',
duration='%sms' % (Ttrans + Tblank),
synapse_id=synAmpaBkg.id)
tpfsPertInh_during = oc.add_transient_poisson_firing_synapse(
nml_doc,
id="tpfsPertInh_during",
average_rate="%s Hz" % (r_bkg_ExtInh + r_stim),
delay='%sms' % (Ttrans + Tblank),
duration='%sms' % (Tstim),
synapse_id=synAmpaBkg.id)
tpfsPertInh_after = oc.add_transient_poisson_firing_synapse(
nml_doc,
id="tpfsPertInh_after",
average_rate="%s Hz" % r_bkg_ExtInh,
delay='%sms' % (Ttrans + Tblank + Tstim),
duration='%sms' % (Tpost),
synapse_id=synAmpaBkg.id)
##### Populations
popExc = oc.add_population_in_rectangular_region(network,
'popExc',
exc_cell_id,
num_exc,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color=exc_color)
from neuroml import Property
popExc.properties.append(Property('type', 'E'))
allExc = [popExc]
if num_exc2 > 0:
popExc2 = oc.add_population_in_rectangular_region(network,
'popExc2',
exc2_cell_id,
num_exc2,
xs,
yDim / 2,
zs,
xDim,
yDimExc2,
zDim,
color=exc2_color)
popExc2.properties.append(Property('type', 'E'))
allExc.append(popExc2)
popInh = oc.add_population_in_rectangular_region(network,
'popInh',
inh_cell_id,
num_inh,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color=inh_color)
popInh.properties.append(Property('type', 'I'))
allInh = [popInh]
if num_inh2 > 0:
popInh2 = oc.add_population_in_rectangular_region(network,
'popInh2',
inh2_cell_id,
num_inh2,
xs,
ys,
zs,
xDim,
yDim,
zDim,
color=inh2_color)
allInh.append(popInh2)
##### Projections
if connections:
weight_expr = 'abs(normal(1,0.5))'
for popEpr in allExc:
for popEpo in allExc:
proj = add_projection(network, "projEE", popEpr, popEpo,
synAmpaEE.id, exc_exc_conn_prob,
global_delay, exc_target_dendrites,
weight_expr)
for popIpo in allInh:
proj = add_projection(network, "projEI", popEpr, popIpo,
synAmpaEI.id, exc_inh_conn_prob,
global_delay, exc_target_dendrites,
weight_expr)
for popIpr in allInh:
for popEpo in allExc:
proj = add_projection(network, "projIE", popIpr, popEpo,
synGabaIE.id, inh_exc_conn_prob,
global_delay, inh_target_dendrites,
weight_expr)
for popIpo in allInh:
proj = add_projection(network, "projII", popIpr, popIpo,
synGabaII.id, inh_inh_conn_prob,
global_delay, inh_target_dendrites,
weight_expr)
elif connections2:
weight_expr = 'abs(normal(1,0.5))'
proj = add_projection(network, "projEE", popExc, popExc, synAmpaEE.id,
exc_exc_conn_prob, global_delay,
exc_target_dendrites, weight_expr)
proj = add_projection(network, "projEI", popExc, popInh, synAmpaEI.id,
exc_inh_conn_prob, global_delay,
exc_target_dendrites, weight_expr)
proj = add_projection(network, "projIE", popInh, popExc, synGabaIE.id,
inh_exc_conn_prob, global_delay,
inh_target_dendrites, weight_expr)
proj = add_projection(network, "projII", popInh, popInh, synGabaII.id,
inh_inh_conn_prob, global_delay,
inh_target_dendrites, weight_expr)
proj = add_projection(network, "projEE2", popExc, popExc2,
synAmpaEE2.id, ee2_conn_prob, global_delay,
exc_target_dendrites, weight_expr)
proj = add_projection(network, "projIE2", popInh, popExc2,
synGabaIE2.id, ie2_conn_prob, global_delay,
inh_target_dendrites, weight_expr)
##### Inputs
oc.add_inputs_to_population(network,
"Stim_E",
popExc,
tpfsExtExc.id,
all_cells=True)
if num_exc2 > 0:
oc.add_inputs_to_population(network,
"Stim_E2",
popExc2,
tpfsExtExc2.id,
all_cells=True)
num_inh_pert = int(popInh.get_size() * fraction_inh_pert)
oc.add_inputs_to_population(network,
"Stim_I_nonpert",
popInh,
tpfsExtInh.id,
all_cells=False,
only_cells=range(num_inh_pert,
popInh.get_size()))
oc.add_inputs_to_population(network,
"Stim_I_pert_before",
popInh,
tpfsPertInh_before.id,
all_cells=False,
only_cells=range(0, num_inh_pert))
oc.add_inputs_to_population(network,
"Stim_I_pert_during",
popInh,
tpfsPertInh_during.id,
all_cells=False,
only_cells=range(0, num_inh_pert))
oc.add_inputs_to_population(network,
"Stim_I_pert_after",
popInh,
tpfsPertInh_after.id,
all_cells=False,
only_cells=range(0, num_inh_pert))
# injecting noise in the soma of detailed neurons to insert some variability
'''oc.add_targeted_inputs_to_population(network,
"PG_noise",
popExc2,
'noisyCurrentSource1', # from ../../../NoisyCurrentSource.xml
segment_group='soma_group',
number_per_cell = 1,
all_cells=True)
oc.add_inputs_to_population(network, "Stim_pre_ExtExc_%s"%popExc.id,
popExc, tpfsExtExc.id,
all_cells=True)
for pop in allExc:
#oc.add_inputs_to_population(network, "Stim_pre_ExtExc_%s"%pop.id,
# pop, tpfsExtExc.id,
# all_cells=True)
oc.add_inputs_to_population(network, "Stim_pre_%s"%pop.id,
pop, tpfsA.id,
all_cells=True)
oc.add_inputs_to_population(network, "Stim_E_%s"%pop.id,
pop, tpfsB.id,
all_cells=True)
for pop in allInh:
num_inh_pert = int(pop.get_size()*fraction_inh_pert)
oc.add_inputs_to_population(network, "Stim_pre_ExtInh_%s"%pop.id,
pop, tpfsExtInh.id,
all_cells=True)
oc.add_inputs_to_population(network, "Stim_pre_%s"%pop.id,
pop, tpfsA.id,
all_cells=True)
oc.add_inputs_to_population(network, "Stim_I_pert_%s"%pop.id,
pop, tpfsC.id,
all_cells=False,
only_cells=range(0,num_inh_pert))
oc.add_inputs_to_population(network, "Stim_I_nonpert_%s"%pop.id,
pop, tpfsB.id,
all_cells=False,
only_cells=range(num_inh_pert,pop.get_size())) '''
save_v = {}
plot_v = {}
# Work in progress...
# General idea: clamp one (or more) exc cell at rev pot of inh syn and see only exc inputs
#
if v_clamp:
levels = {'IPSC': synAmpaEE.erev, 'EPSC': synGabaIE.erev}
for l in levels:
cell_index = levels.keys().index(l)
pop = 'popExc2'
plot = 'IClamp_i_%s' % (l)
for seg_id in [0, 2953,
1406]: # 2953: end of axon; 1406 on dendrite
clamp_id = "vclamp_cell%s_seg%s_%s" % (cell_index, seg_id, l)
v_clamped = levels[l]
vc = oc.add_voltage_clamp_triple(
nml_doc,
id=clamp_id,
delay='0ms',
duration='%sms' % duration,
conditioning_voltage=v_clamped,
testing_voltage=v_clamped,
return_voltage=v_clamped,
simple_series_resistance="1e2ohm",
active="1")
vc_dat_file = 'v_clamps_i_seg%s_%s.%s.dat' % (seg_id, l,
simulation_seed)
seg_file = '%s_seg%s_%s_v.dat' % (pop, seg_id, l)
save_v[vc_dat_file] = []
plot_v[plot] = []
oc.add_inputs_to_population(network,
"vclamp_seg%s_%s" % (seg_id, l),
network.get_by_id(pop),
vc.id,
all_cells=False,
only_cells=[cell_index],
segment_ids=[seg_id])
# record at seg
q = '%s/%s/%s/%s/%s/i' % (pop, cell_index,
network.get_by_id(pop).component,
seg_id, clamp_id)
save_v[vc_dat_file].append(q)
plot_v[plot].append(q)
if seg_id != 0:
save_v[seg_file] = []
q = '%s/%s/%s/%s/v' % (pop, cell_index,
network.get_by_id(pop).component,
seg_id)
save_v[seg_file].append(q)
##### Save NeuroML and LEMS Simulation files
nml_file_name = '%s.net.%s' % (network.id,
'nml.h5' if format == 'hdf5' else 'nml')
oc.save_network(nml_doc,
nml_file_name,
validate=(format == 'xml'),
format=format,
target_dir=target_dir)
print("Saved to: %s" % nml_file_name)
if num_exc > 0:
exc_traces = '%s_%s_v.dat' % (network.id, popExc.id)
save_v[exc_traces] = []
plot_v[popExc.id] = []
if num_inh > 0:
inh_traces = '%s_%s_v.dat' % (network.id, popInh.id)
save_v[inh_traces] = []
plot_v[popInh.id] = []
if num_exc2 > 0:
exc2_traces = '%s_%s_v.dat' % (network.id, popExc2.id)
save_v[exc2_traces] = []
plot_v[popExc2.id] = []
if num_inh2 > 0:
inh2_traces = '%s_%s_v.dat' % (network.id, popInh2.id)
save_v[inh2_traces] = []
plot_v[popInh2.id] = []
for i in range(min(max_in_pop_to_plot_and_save, num_exc)):
plot_v[popExc.id].append("%s/%i/%s/v" %
(popExc.id, i, popExc.component))
save_v[exc_traces].append("%s/%i/%s/v" %
(popExc.id, i, popExc.component))
for i in range(min(max_in_pop_to_plot_and_save, num_exc2)):
plot_v[popExc2.id].append("%s/%i/%s/v" %
(popExc2.id, i, popExc2.component))
save_v[exc2_traces].append("%s/%i/%s/v" %
(popExc2.id, i, popExc2.component))
for i in range(min(max_in_pop_to_plot_and_save, num_inh)):
plot_v[popInh.id].append("%s/%i/%s/v" %
(popInh.id, i, popInh.component))
save_v[inh_traces].append("%s/%i/%s/v" %
(popInh.id, i, popInh.component))
for i in range(min(max_in_pop_to_plot_and_save, num_inh2)):
plot_v[popInh2.id].append("%s/%i/%s/v" %
(popInh2.id, i, popInh2.component))
save_v[inh2_traces].append("%s/%i/%s/v" %
(popInh2.id, i, popInh2.component))
gen_spike_saves_for_all_somas = True
lems_file_name, lems_sim = oc.generate_lems_simulation(
nml_doc,
network,
target_dir + nml_file_name,
duration=duration,
dt=dt,
gen_plots_for_all_v=False,
gen_plots_for_quantities=plot_v,
gen_saves_for_all_v=False,
gen_saves_for_quantities=save_v,
gen_spike_saves_for_all_somas=gen_spike_saves_for_all_somas,
target_dir=target_dir,
include_extra_lems_files=['./NoisyCurrentSource.xml'],
report_file_name='report.txt',
simulation_seed=simulation_seed)
if run_in_simulator:
print("Running %s for %sms in %s" %
(lems_file_name, duration, run_in_simulator))
traces, events = oc.simulate_network(lems_file_name,
run_in_simulator,
max_memory='4000M',
nogui=True,
load_saved_data=True,
reload_events=True,
plot=False,
verbose=True,
num_processors=num_processors)
print("Reloaded traces: %s" % traces.keys())
#print("Reloaded events: %s"%events.keys())
use_events_for_rates = False
exc_rate = 0
inh_rate = 0
if use_events_for_rates:
if (run_in_simulator == 'jNeuroML_NetPyNE'):
raise (
'Saving of spikes (and so calculation of rates) not yet supported in jNeuroML_NetPyNE'
)
for ek in events.keys():
rate = 1000 * len(events[ek]) / float(duration)
print("Cell %s has a rate %s Hz" % (ek, rate))
if 'popExc' in ek:
exc_rate += rate / num_exc
if 'popInh' in ek:
inh_rate += rate / num_inh
else:
tot_exc_rate = 0
exc_cells = 0
tot_inh_rate = 0
inh_cells = 0
tt = [t * 1000 for t in traces['t']]
for tk in traces.keys():
if tk != 't':
rate = get_rate_from_trace(tt,
[v * 1000 for v in traces[tk]])
print("Cell %s has rate %s Hz" % (tk, rate))
if 'popExc' in tk:
tot_exc_rate += rate
exc_cells += 1
if 'popInh' in tk:
tot_inh_rate += rate
inh_cells += 1
exc_rate = tot_exc_rate / exc_cells
inh_rate = tot_inh_rate / inh_cells
print("Run %s: Exc rate: %s Hz; Inh rate %s Hz" %
(reference, exc_rate, inh_rate))
return exc_rate, inh_rate, traces
return nml_doc, nml_file_name, lems_file_name
