def run_hyperpolarizing_learning_subjects(subj_ids, stim_types, coherence_levels, trials_per_condition):
(p_dcs, i_dcs) = (-stim_intensity_max, 0.5*stim_intensity_max)
conditions= OrderedDict({'training': (simulation_params(ntrials=trials_per_condition, plasticity=True, p_dcs=p_dcs, i_dcs=i_dcs, dcs_start_time=0*second, dcs_end_time=4*second), True, coherence_levels)})
conditions['testing'] = (simulation_params(ntrials=trials_per_condition), True, coherence_levels)
print conditions.keys()
print p_dcs, i_dcs
# Run subjects
run_virtual_subjects(subj_ids, conditions,
'/home/jeff/projects/pySBI/data/stdp_maxstim_2_plasticity_14_40_fixed/final/1/hyperpolarizing/final_test_final',
'/home/jeff/projects/pySBI/data/rerw/fitted_behavioral_params.h5')
def __init__(self, subj_id, wta_params=default_params(), pyr_params=pyr_params(), inh_params=inh_params(),
plasticity_params=plasticity_params(), sim_params=simulation_params()):
self.subj_id = subj_id
self.wta_params = wta_params
self.pyr_params = pyr_params
self.inh_params = inh_params
self.plasticity_params = plasticity_params
self.sim_params = sim_params
self.simulation_clock = Clock(dt=self.sim_params.dt)
self.input_update_clock = Clock(dt=1 / (self.wta_params.refresh_rate / Hz) * second)
self.background_input = PoissonGroup(self.wta_params.background_input_size,
rates=self.wta_params.background_freq, clock=self.simulation_clock)
self.task_inputs = []
for i in range(self.wta_params.num_groups):
self.task_inputs.append(PoissonGroup(self.wta_params.task_input_size,
rates=self.wta_params.task_input_resting_rate, clock=self.simulation_clock))
# Create WTA network
self.wta_network = WTANetworkGroup(params=self.wta_params, background_input=self.background_input,
task_inputs=self.task_inputs, pyr_params=self.pyr_params, inh_params=self.inh_params,
plasticity_params=self.plasticity_params, clock=self.simulation_clock)
# Create network monitor
self.wta_monitor = WTAMonitor(self.wta_network, None, None, self.sim_params, record_lfp=False,
record_voxel=False, record_neuron_state=False, record_spikes=False,
record_firing_rate=True, record_inputs=True, record_connections=None,
save_summary_only=False, clock=self.simulation_clock)
# Create Brian network and reset clock
self.net = Network(self.background_input, self.task_inputs, self.wta_network,
self.wta_network.connections.values(), self.wta_monitor.monitors.values())
def post_wta_jobs(nodes,
p_b_e_range,
p_x_e_range,
p_e_e_range,
p_e_i_range,
p_i_i_range,
p_i_e_range,
num_trials,
muscimol_amount=0 * nS,
injection_site=0,
start_nodes=True):
sim_params = simulation_params()
sim_params.muscimol_amount = muscimol_amount
sim_params.injection_site = injection_site
input_sum = 40.0
launcher = Launcher(nodes)
if start_nodes:
launcher.set_application_script(
os.path.join(SRC_DIR, 'sh/ezrcluster-application-script.sh'))
launcher.start_nodes()
contrast_range = [0.0, 0.0625, 0.125, 0.25, 0.5, 1.0]
for p_b_e in p_b_e_range:
for p_x_e in p_x_e_range:
for p_e_e in p_e_e_range:
for p_e_i in p_e_i_range:
for p_i_i in p_i_i_range:
for p_i_e in p_i_e_range:
wta_params = default_params()
wta_params.p_b_e = p_b_e
wta_params.p_x_e = p_x_e
wta_params.p_e_e = p_e_e
wta_params.p_e_i = p_e_i
wta_params.p_i_i = p_i_i
wta_params.p_i_e = p_i_e
for i, contrast in enumerate(contrast_range):
inputs = np.zeros(2)
inputs[0] = (input_sum * (contrast + 1.0) /
2.0)
inputs[1] = input_sum - inputs[0]
for t in range(num_trials):
np.random.shuffle(inputs)
cmds, log_file_template, out_file = get_wta_cmds(
wta_params,
inputs,
sim_params,
contrast,
t,
record_lfp=True,
record_voxel=True,
record_neuron_state=False,
record_firing_rate=True,
record_spikes=True)
launcher.add_job(
cmds,
log_file_template=log_file_template,
output_file=out_file)
def launch_virtual_subject_processes(nodes, mu_0, virtual_subj_ids, behavioral_param_file, trials, stim_conditions,
start_nodes=True):
"""
nodes = nodes to run simulation on
data_dir = directory containing subject data
num_real_subjects = number of real subjects
num_virtual_subjects = number of virtual subjects to run
behavioral_param_file = file containing subject fitted behavioral parameters
start_nodes = whether or not to start nodes
"""
# Setup launcher
launcher=Launcher(nodes)
wta_params=default_params()
wta_params.mu_0=mu_0
wta_params.p_a=wta_params.mu_0/100.0
wta_params.p_b=wta_params.p_a
# Get subject alpha and beta values
f = h5py.File(behavioral_param_file)
control_group=f['control']
alpha_vals=np.array(control_group['alpha'])
beta_vals=np.array(control_group['beta'])
# For each virtual subject
for virtual_subj_id in virtual_subj_ids:
# Sample beta from subject distribution - don't use subjects with high alpha
beta_hist,beta_bins=np.histogram(beta_vals[np.where(alpha_vals<.99)[0]], density=True)
bin_width=beta_bins[1]-beta_bins[0]
beta_bin=np.random.choice(beta_bins[:-1], p=beta_hist*bin_width)
beta=beta_bin+np.random.rand()*bin_width
wta_params.background_freq=(beta-161.08)/-.17
contrast_range=[0.0, .032, .064, .096, .128, .256, .512]
for i,contrast in enumerate(contrast_range):
inputs=np.array([wta_params.mu_0+wta_params.p_a*contrast*100.0,
wta_params.mu_0-wta_params.p_b*contrast*100.0])
for t in range(trials):
np.random.shuffle(inputs)
for stim_condition,stim_values in stim_conditions.iteritems():
sim_params=simulation_params()
sim_params.p_dcs=stim_values[0]
sim_params.i_dcs=stim_values[1]
cmds,log_file_template,out_file=get_wta_cmds(wta_params, inputs, sim_params, contrast, t,
record_lfp=True, record_voxel=True, record_neuron_state=False, record_firing_rate=True,
record_spikes=True, save_summary_only=False,
e_desc='virtual_subject.%d.%s' % (virtual_subj_id,stim_condition))
launcher.add_batch_job(cmds, log_file_template=log_file_template, output_file=out_file)
launcher.post_jobs()
if start_nodes:
launcher.set_application_script(os.path.join(SRC_DIR, 'sh/ezrcluster-application-script.sh'))
launcher.start_nodes()
def run_nostim_training_subjects(subj_ids, stim_types, coherence_levels, trials_per_condition):
"""
Run subjects with no stimulation during training - runs baseline without stim, training without stim, and test without stim
nsubjects = number of subjects to simulation
coherence_levels = number of coherence levels to use
"""
conditions = OrderedDict({'baseline': (simulation_params(ntrials=trials_per_condition), True, coherence_levels)})
# conditions = OrderedDict({'training': (simulation_params(ntrials=trials_per_condition, plasticity=True), True, coherence_levels)})
conditions['training']= (simulation_params(ntrials=trials_per_condition, plasticity=True), True, coherence_levels)
# conditions = OrderedDict({'training': (simulation_params(ntrials=trials_per_condition, plasticity=True), True, coherence_levels)})
conditions['testing'] = (simulation_params(ntrials=trials_per_condition), True, coherence_levels)
# run baseline with stim
# for stim in stim_types:
# p_dcs, i_dcs = stim_types.get(stim)
# conditions['baseline_%s' % stim] = (simulation_params(ntrials=trials_per_condition, plasticity=False, p_dcs=p_dcs, i_dcs=i_dcs, dcs_start_time=0*second, dcs_end_time=4*second), True, coherence_levels)
print conditions.keys()
# Run subject
run_virtual_subjects(subj_ids, conditions, '/home/jeff/projects/pySBI/data/stdp_maxstim_2_plasticity_14_40_fixed/final/6/control/final_test_final',
'/home/jeff/projects/pySBI/data/rerw/fitted_behavioral_params.h5')
def post_wta_jobs(nodes, p_b_e_range, p_x_e_range, p_e_e_range, p_e_i_range, p_i_i_range, p_i_e_range, num_trials,
muscimol_amount=0*nS, injection_site=0, start_nodes=True):
sim_params=simulation_params()
sim_params.muscimol_amount=muscimol_amount
sim_params.injection_site=injection_site
input_sum=40.0
launcher=Launcher(nodes)
if start_nodes:
launcher.set_application_script(os.path.join(SRC_DIR, 'sh/ezrcluster-application-script.sh'))
launcher.start_nodes()
contrast_range=[0.0, 0.0625, 0.125, 0.25, 0.5, 1.0]
for p_b_e in p_b_e_range:
for p_x_e in p_x_e_range:
for p_e_e in p_e_e_range:
for p_e_i in p_e_i_range:
for p_i_i in p_i_i_range:
for p_i_e in p_i_e_range:
wta_params=default_params()
wta_params.p_b_e=p_b_e
wta_params.p_x_e=p_x_e
wta_params.p_e_e=p_e_e
wta_params.p_e_i=p_e_i
wta_params.p_i_i=p_i_i
wta_params.p_i_e=p_i_e
for i,contrast in enumerate(contrast_range):
inputs=np.zeros(2)
inputs[0]=(input_sum*(contrast+1.0)/2.0)
inputs[1]=input_sum-inputs[0]
for t in range(num_trials):
np.random.shuffle(inputs)
cmds,log_file_template,out_file=get_wta_cmds(wta_params, inputs, sim_params,
contrast, t, record_lfp=True, record_voxel=True, record_neuron_state=False,
record_firing_rate=True, record_spikes=True)
launcher.add_job(cmds, log_file_template=log_file_template, output_file=out_file)
def run_rl_simulation(mat_file, alpha=0.4, beta=5.0, background_freq=None, p_dcs=0*pA, i_dcs=0*pA, dcs_start_time=0*ms,
output_file=None):
mat = scipy.io.loadmat(mat_file)
prob_idx=-1
mags_idx=-1
for idx,(dtype,o) in enumerate(mat['store']['dat'][0][0].dtype.descr):
if dtype=='probswalk':
prob_idx=idx
elif dtype=='mags':
mags_idx=idx
prob_walk=mat['store']['dat'][0][0][0][0][prob_idx]
mags=mat['store']['dat'][0][0][0][0][mags_idx]
prob_walk=prob_walk.astype(np.float32, copy=False)
mags=mags.astype(np.float32, copy=False)
mags /= 100.0
wta_params=default_params()
wta_params.input_var=0*Hz
sim_params=simulation_params()
sim_params.p_dcs=p_dcs
sim_params.i_dcs=i_dcs
sim_params.dcs_start_time=dcs_start_time
exp_rew=np.array([0.5, 0.5])
if background_freq is None:
background_freq=(beta-161.08)/-.17
wta_params.background_freq=background_freq
trials=prob_walk.shape[1]
sim_params.ntrials=trials
vals=np.zeros(prob_walk.shape)
choice=np.zeros(trials)
rew=np.zeros(trials)
rts=np.zeros(trials)
inputs=np.zeros(prob_walk.shape)
if output_file is not None:
f = h5py.File(output_file, 'w')
f.attrs['alpha']=alpha
f.attrs['beta']=beta
f.attrs['mat_file']=mat_file
f_sim_params=f.create_group('sim_params')
for attr, value in sim_params.iteritems():
f_sim_params.attrs[attr] = value
f_network_params=f.create_group('network_params')
for attr, value in wta_params.iteritems():
f_network_params.attrs[attr] = value
f_pyr_params=f.create_group('pyr_params')
for attr, value in pyr_params.iteritems():
f_pyr_params.attrs[attr] = value
f_inh_params=f.create_group('inh_params')
for attr, value in inh_params.iteritems():
f_inh_params.attrs[attr] = value
for trial in range(sim_params.ntrials):
print('Trial %d' % trial)
vals[:,trial]=exp_rew
ev=vals[:,trial]*mags[:,trial]
inputs[0,trial]=ev[0]
inputs[1,trial]=ev[1]
inputs[:,trial]=40.0+40.0*inputs[:,trial]
trial_monitor=run_wta(wta_params, inputs[:,trial], sim_params, record_lfp=False, record_voxel=False,
record_neuron_state=False, record_spikes=True, record_firing_rate=True, record_inputs=False,
plot_output=False)
e_rates = []
for i in range(wta_params.num_groups):
e_rates.append(trial_monitor.monitors['excitatory_rate_%d' % i].smooth_rate(width=5 * ms, filter='gaussian'))
i_rates = [trial_monitor.monitors['inhibitory_rate'].smooth_rate(width=5 * ms, filter='gaussian')]
if output_file is not None:
trial_group=f.create_group('trial %d' % trial)
trial_group['e_rates'] = np.array(e_rates)
trial_group['i_rates'] = np.array(i_rates)
rt,decision_idx=get_response_time(e_rates, sim_params.stim_start_time, sim_params.stim_end_time,
upper_threshold=wta_params.resp_threshold, lower_threshold=None, dt=sim_params.dt)
reward=0.0
if decision_idx>=0 and np.random.random()<=prob_walk[decision_idx,trial]:
reward=1.0
exp_rew[decision_idx]=(1.0-alpha)*exp_rew[decision_idx]+alpha*reward
choice[trial]=decision_idx
rts[trial]=rt
rew[trial]=reward
param_ests,prop_correct=fit_behavior(prob_walk, mags, rew, choice)
if output_file is not None:
f.attrs['est_alpha']=param_ests[0]
f.attrs['est_beta']=param_ests[1]
f.attrs['prop_correct']=prop_correct
f['prob_walk']=prob_walk
f['mags']=mags
f['rew']=rew
f['choice']=choice
f['vals']=vals
f['inputs']=inputs
f['rts']=rts
f.close()
session_monitor.plot()
else:
subject.wta_monitor.plot()
plt.show()
if __name__ == "__main__":
# Trials per condition
trials_per_condition = 100
# Max stimulation intensity
stim_intensity_max = 0.75 * pA
# Stimulation conditions
conditions = {
"control": simulation_params(
ntrials=trials_per_condition,
trial_duration=3 * second,
stim_start_time=1 * second,
stim_end_time=2 * second,
),
"depolarizing": simulation_params(
ntrials=trials_per_condition,
trial_duration=3 * second,
stim_start_time=1 * second,
stim_end_time=2 * second,
p_dcs=stim_intensity_max,
i_dcs=-0.5 * stim_intensity_max,
dcs_start_time=0 * second,
dcs_end_time=3 * second,
),
"hyperpolarizing": simulation_params(
ntrials=trials_per_condition,
trial_duration=3 * second,
def launch_control_virtual_subject_processes(nodes, mu_0, virtual_subj_ids, behavioral_param_file, trials,
stim_gains=[8,6,4,2,1,0.5,0.25], start_nodes=True):
"""
Launch stimulation intensity simulations with DCS applied only to pyramidal population
nodes = nodes to run simulation on
data_dir = directory containing subject data
num_real_subjects = number of real subjects
num_virtual_subjects = number of virtual subjects to run
behavioral_param_file = file containing subject fitted behavioral parameters
start_nodes = whether or not to start nodes
"""
# Setup launcher
launcher=Launcher(nodes)
# Get subject alpha and beta values
f = h5py.File(behavioral_param_file)
control_group=f['control']
alpha_vals=np.array(control_group['alpha'])
beta_vals=np.array(control_group['beta'])
# For each virtual subject
for virtual_subj_id in virtual_subj_ids:
wta_params=default_params()
wta_params.mu_0=mu_0
wta_params.p_a=mu_0/100.0
wta_params.p_b=wta_params.p_a
# Sample beta from subject distribution - don't use subjects with high alpha
beta_hist,beta_bins=np.histogram(beta_vals[np.where(alpha_vals<.99)[0]], density=True)
bin_width=beta_bins[1]-beta_bins[0]
beta_bin=np.random.choice(beta_bins[:-1], p=beta_hist*bin_width)
beta=beta_bin+np.random.rand()*bin_width
wta_params.background_freq=(beta-161.08)/-.17
contrast_range=[0.0, .032, .064, .128, .256, .512]
for i,contrast in enumerate(contrast_range):
inputs=np.zeros(2)
inputs[0]=wta_params.mu_0+wta_params.p_a*contrast*100.0
inputs[1]=wta_params.mu_0-wta_params.p_b*contrast*100.0
for t in range(trials):
np.random.shuffle(inputs)
for idx, stim_gain in enumerate(stim_gains):
sim_params=simulation_params()
sim_params.p_dcs=stim_gain*pA
cmds,log_file_template,out_file=get_wta_cmds(wta_params, inputs, sim_params, contrast, t,
record_lfp=True, record_voxel=True, record_neuron_state=False, record_firing_rate=True,
record_spikes=True, save_summary_only=False,
e_desc='virtual_subject.%d.anode' % virtual_subj_id)
launcher.add_batch_job(cmds, log_file_template=log_file_template, output_file=out_file)
sim_params=simulation_params()
sim_params.p_dcs=-stim_gain*pA
cmds,log_file_template,out_file=get_wta_cmds(wta_params, inputs, sim_params, contrast, t,
record_lfp=True, record_voxel=True, record_neuron_state=False, record_firing_rate=True,
record_spikes=True, save_summary_only=False,
e_desc='virtual_subject.%d.cathode' % virtual_subj_id)
launcher.add_batch_job(cmds, log_file_template=log_file_template, output_file=out_file)
if idx==0:
sim_params=simulation_params()
cmds,log_file_template,out_file=get_wta_cmds(wta_params, inputs, sim_params, contrast, t,
record_lfp=True, record_voxel=True, record_neuron_state=False, record_firing_rate=True,
record_spikes=True, save_summary_only=False,
e_desc='virtual_subject.%d.control' % virtual_subj_id)
launcher.add_batch_job(cmds, log_file_template=log_file_template, output_file=out_file)
launcher.post_jobs()
if start_nodes:
launcher.set_application_script(os.path.join(SRC_DIR, 'sh/ezrcluster-application-script.sh'))
launcher.start_nodes()
