def load_spike_times_spontaneous():
bcs = iaf.get_continue_bcs(n=40)
print bcs
sim = bluepy.Simulation(bcs[0])
gids = list(sim.get_circuit_target())
df = load_spike_times(bcs, gids)
return df
def spike_count_difference():
n_include = 20
bcs_0 = iaf.get_continue_bcs(params='')[:n_include]
bcs_1 = iaf.get_change_bcs(params='y')[:n_include]
bcs_2 = iaf.get_change_bcs(params='e')[:n_include]
bcs_3 = iaf.get_change_bcs(params='abcd')[:n_include]
bcs_4 = iaf.get_change_bcs(params='abcdy')[:n_include]
bcs_list = [bcs_1, bcs_2, bcs_3, bcs_4]
bins=np.linspace(0, 50, 11)
differences = np.zeros((4, len(bcs_0), bins.size - 1))
for i, bcs in enumerate(bcs_list):
differences[i, :, :] = get_differences(bcs_0, bcs, bins=bins)
means = differences.mean(axis=1)
errs = np.apply_along_axis(iaf.mean_confidence_interval, 1, differences)
fig, axs = plt.subplots(2)
ax = axs[0]
xs = bins[:-1] + (bins[1] - bins[0])/2.0
ax.plot(xs, np.zeros(xs.size), color='red')
ax.errorbar(xs, means[0], errs[0])
ax.errorbar(xs + 0.2, means[1], errs[1])
ax = axs[1]
xs = bins[:-1] + (bins[1] - bins[0])/2.0
ax.plot(xs, np.zeros(xs.size), color='red')
ax.errorbar(xs, means[2], errs[2])
ax.errorbar(xs + 0.2, means[3], errs[3])
plt.savefig('figures/perturbation_fr.pdf')
def plot_count_variance():
"""
:return:
"""
circuit = bluepy.Simulation(iaf.get_continue_bcs(n=2)[0]).circuit
df_neurons = circuit.v2.cells({Cell.HYPERCOLUMN: 2})
fig, axs = plt.subplots(6, 4, figsize=(10, 12))
mean_fr = []
for i, n in enumerate([40, 20, 10, 4, 2, 1]):
df_counts = get_spike_counts_spontaneous_binned_concatenated(np.linspace(0, 2000, n+1))
df_counts_exc = df_counts[df_neurons['synapse_class'] == 'EXC']
df_counts_inh = df_counts[df_neurons['synapse_class'] == 'INH']
mean_fr.append(df_counts['mean'].loc[84815])
ax = axs[0 + i, 0]
ax.scatter(df_counts_exc['mean'], df_counts_exc['variance'], marker='.', color='red', s=3, rasterized=True)
xs = np.linspace(0, df_counts_exc['mean'].max(), 1000)
ax.plot(xs, xs, '--', color='black')
ax.plot(xs, (xs % 1) * (1 - (xs % 1)), '-', color='black')
ax.set_ylim([0, xs[-1]])
ax = axs[0 + i, 2]
ax.hist(df_counts_exc['ff'], bins=np.linspace(0, 2, 20), histtype='step', color='red')
#ax.set_xlim([0, 10])
ax = axs[0 + i, 1]
ax.scatter(df_counts_inh['mean'], df_counts_inh['variance'], marker='.', color='blue', s=3, rasterized=True)
xs = np.linspace(0, df_counts_inh['mean'].max(), 1000)
ax.plot(xs, xs, '--', color='black')
ax.plot(xs, (xs % 1) * (1 - (xs % 1)), '-', color='black')
ax.set_ylim([0, xs[-1]])
ax = axs[0 + i, 3]
ax.hist(df_counts_inh['ff'], bins=np.linspace(0, 2, 20), histtype='step', color='blue')
#ax.set_xlim([0, 80])
print mean_fr
for ax in axs[:, :2].flatten():
ax.set_xlabel('Spike count')
ax.set_ylabel('Variance')
for ax in axs[:, 2:].flatten():
ax.set_xlabel('Fano factor')
ax.set_ylabel('Neurons')
for k, ax in enumerate(axs[:, 0]):
ax.set_title(['50 ms', '100 ms', '200 ms', '500 ms', '1000 ms', '2000 ms'][k])
plt.tight_layout()
plt.savefig('figures_poisson/spont_variance_poisson.pdf', dpi=300)
def plot_count_variance_exp_25_summary():
bins = np.linspace(1000, 7000, 13)
print bins
circuit = bluepy.Simulation(iaf.get_continue_bcs(n=2)[0]).circuit
# df_neurons = circuit.v2.cells({Cell.HYPERCOLUMN: 2})
df_neurons = get_selected_L456_gids()
fig, axs = plt.subplots(5, len(bins) - 1, figsize=(20, 6))
for k, variable in enumerate([False, True]):
df_counts_list = [get_spike_counts_evoked(t_start=bins[j], t_end=bins[j + 1],
variable=variable) for j in range(len(bins) - 1)]
df_counts_exc_list = []
for i, df_counts in enumerate(df_counts_list):
df_counts_exc = df_counts.loc[df_neurons.index.unique()][df_neurons['synapse_class'] == 'EXC']
df_counts_exc_list.append(df_counts_exc)
df_counts_inh = df_counts.loc[df_neurons.index.unique()][df_neurons['synapse_class'] == 'INH']
ax = axs[0 + k * 2, i]
ax.scatter(df_counts_exc['mean'], df_counts_exc['variance'], marker='.', color='red', s=3, rasterized=True)
xs = np.linspace(0, df_counts_exc['mean'].max(), 1000)
ax.plot(xs, xs, '--', color='black')
ax.plot(xs, (xs % 1) * (1 - (xs % 1)), '-', color='black')
ax.set_xlim([0, xs.max()])
ax.set_ylim([0, xs.max()])
ax = axs[1 + k * 2, i]
ax.scatter(df_counts_inh['mean'], df_counts_inh['variance'], marker='.', color='blue', s=3, rasterized=True)
xs = np.linspace(0, df_counts_inh['mean'].max(), 1000)
ax.plot(xs, xs, '--', color='black')
ax.plot(xs, (xs % 1) * (1 - (xs % 1)), '-', color='black')
ax.set_xlim([0, xs.max()])
ax.set_ylim([0, xs.max()])
df_counts_list_vpm = [get_spike_counts_vpm(t_start=bins[j], t_end=bins[j + 1]) for j in range(2, len(bins) - 1)]
for i, df_counts in enumerate(df_counts_list_vpm):
ax = axs[-1, i+2]
ax.scatter(df_counts['mean'], df_counts['variance'], marker='.', color='green', s=3, rasterized=True)
xs = np.linspace(0, df_counts['mean'].max(), 1000)
ax.plot(xs, xs, '--', color='black')
ax.plot(xs, (xs % 1) * (1 - (xs % 1)), '-', color='black')
ax.set_xlim([0, xs.max()])
ax.set_ylim([0, xs.max()])
plt.savefig('figures_poisson/exp_25_variance_poisson_summary.pdf', dpi=300)
def compute_spike_counts(df, n=30, vpm=False):
bcs = iaf.get_continue_bcs(n=2)
sim = bluepy.Simulation(bcs[0])
if not vpm:
gids = np.sort(np.array(list(sim.get_circuit_target())))
else:
gids = np.arange(221042, 221042 + 310)
spikes_df = df.drop(columns=['time'])
counts_all = np.zeros((gids.size, n), dtype=np.float64)
for i in range(n):
x = spikes_df[spikes_df['trial'] == i]
counts_all[:, i] = np.bincount(np.array(x.index).astype(int), minlength=gids.max()+1)[-gids.size:]
means = counts_all.mean(axis=1)
vars = np.var(counts_all, ddof=1, axis=1)
ffs = vars/means
ffs[means == 0] = 0
#print means
df_counts = pd.DataFrame({'mean':means, 'variance':vars, 'ff':ffs}, index=gids)
return df_counts