def raster_from_epoch(signals, epoch, occurrences=0, channels=0,
xlabel='Time', ylabel='Value',
linestyle='-', linewidth=1,
ax=None, title=None):
"""TODO: doc"""
raise NotImplementedError
if occurrences is None:
return
occurrences = pad_to_signals(signals, occurrences)
channels = pad_to_signals(signals, channels)
legend = [s.name for s in signals]
times = []
values = []
for s, o, c in zip(signals, occurrences, channels):
# Get occurrences x chans x time
extracted = s.extract_epoch(epoch)
# Get values from specified occurrence and channel
value_vector = extracted[o][c]
# Convert bins to time (relative to start of epoch)
# TODO: want this to be absolute time relative to start of data?
time_vector = np.arange(0, len(value_vector)) / s.fs
times.append(time_vector)
values.append(value_vector)
plot_timeseries(times, values, xlabel, ylabel, legend=legend,
linestyle=linestyle, linewidth=linewidth,
ax=ax, title=title)
def raster_from_vectors(vectors, xlabel='Time', ylabel='Value', fs=None,
linestyle='-', linewidth=1, legend=None,
ax=None, title=None):
"""TODO: doc"""
raise NotImplementedError
times = []
values = []
for v in vectors:
values.append(v)
if fs is None:
times.append(np.arange(0, len(v)))
else:
times.append(np.arange(0, len(v))/fs)
plot_timeseries(times, values, xlabel, ylabel, legend=legend,
linestyle=linestyle, linewidth=linewidth,
ax=ax, title=title)
def strf_timeseries(modelspec,
ax=None,
clim=None,
show_factorized=True,
show_fir_only=True,
title=None,
fs=1,
chans=None):
"""
chans: list
if not None, label each row of the strf with the corresponding
channel name
"""
wcc = _get_wc_coefficients(modelspec)
firc = _get_fir_coefficients(modelspec)
if wcc is None and firc is None:
log.warn('Unable to generate STRF.')
return
elif show_fir_only or (wcc is None):
strf = np.array(firc)
elif wcc is not None and firc is None:
strf = np.array(wcc).T
else:
wc_coefs = np.array(wcc).T
fir_coefs = np.array(firc)
if wc_coefs.shape[1] == fir_coefs.shape[0]:
strf = wc_coefs @ fir_coefs
else:
strf = fir_coefs
times = np.arange(strf.shape[1]) / fs
plot_timeseries([times], [strf.T],
xlabel='Time lag',
ylabel='Gain',
legend=chans,
linestyle='-',
linewidth=1,
ax=ax,
title=title)
plt.plot(times[[0, len(times) - 1]],
np.array([0, 0]),
linewidth=0.5,
color='gray')
def raster_from_signals(signals, channels=0, xlabel='Time', ylabel='Value',
linestyle='-', linewidth=1,
ax=None, title=None):
"""TODO: doc"""
raise NotImplementedError
channels = pad_to_signals(signals, channels)
times = []
values = []
legend = []
for s, c in zip(signals, channels):
# Get values from specified channel
value_vector = s.as_continuous()[c]
# Convert indices to absolute time based on sampling frequency
time_vector = np.arange(0, len(value_vector)) / s.fs
times.append(time_vector)
values.append(value_vector)
legend.append(s.name+' '+s.chans[c])
plot_timeseries(times, values, xlabel, ylabel, legend=legend,
linestyle=linestyle, linewidth=linewidth,
ax=ax, title=title)
def psth_from_raster(times, values, xlabel='Time', ylabel='Value',
legend=None, linestyle='-', linewidth=1,
ax=None, title=None, facecolor='lightblue',
binsize=1):
if binsize > 1:
x = np.reshape(values, [values.shape[0], -1, binsize])
x = np.nanmean(x, axis=2)
t = times[int(binsize/2)::binsize]
else:
x = values
t = times
m = np.nanmean(x, axis=0)
e = np.nanstd(x, axis=0) / np.sqrt(np.sum(np.isfinite(x[:, 0])))
if ax is not None:
plt.sca(ax)
plt.fill_between(t, m-e, m+e, facecolor=facecolor)
plot_timeseries([t], [m], xlabel=xlabel, ylabel=ylabel,
legend=legend, linestyle=linestyle,
linewidth=linewidth, ax=ax, title=title)
def strf_timeseries(modelspec,
ax=None,
show_factorized=True,
show_fir_only=True,
title=None,
fs=1,
chans=None,
colors=None,
**options):
"""
chans: list
if not None, label each row of the strf with the corresponding
channel name
"""
wcc = _get_wc_coefficients(modelspec)
firc = _get_fir_coefficients(modelspec)
if wcc is None and firc is None:
log.warn('Unable to generate STRF.')
return
elif show_fir_only or (wcc is None):
strf = np.array(firc)
elif wcc is not None and firc is None:
strf = np.array(wcc).T
else:
wc_coefs = np.array(wcc).T
fir_coefs = np.array(firc)
if wc_coefs.shape[1] == fir_coefs.shape[0]:
strf = wc_coefs @ fir_coefs
else:
strf = fir_coefs
times = [np.arange(strf.shape[1]) / fs] * strf.shape[0]
filters = [strf[i] for i in range(strf.shape[0])]
if colors is None:
if strf.shape[0] == 1:
colors = [[0, 0, 0]]
elif strf.shape[0] == 2:
colors = [[254 / 255, 15 / 255, 6 / 255],
[129 / 255, 201 / 255, 224 / 255]]
elif strf.shape[0] == 3:
colors = [[254 / 255, 15 / 255, 6 / 255],
[217 / 255, 217 / 255, 217 / 255],
[129 / 255, 201 / 255, 224 / 255]]
elif strf.shape[0] > 3:
colors = [[254 / 255, 15 / 255, 6 / 255],
[217 / 255, 217 / 255, 217 / 255],
[129 / 255, 201 / 255, 224 / 255],
[128 / 255, 128 / 255, 128 / 255],
[32 / 255, 32 / 255, 32 / 255]]
#import pdb
#pdb.set_trace()
_, strf_h = plot_timeseries(times,
filters,
xlabel='Time lag',
ylabel='Gain',
legend=chans,
linestyle='-',
linewidth=1,
ax=ax,
title=title,
colors=colors)
plt.plot(times[0][[0, len(times[0]) - 1]],
np.array([0, 0]),
linewidth=0.5,
color='gray')
if show_factorized and not show_fir_only:
wcN = wcc.shape[0]
ax.set_prop_cycle(None)
_, fir_h = plot_timeseries([times], [firc.T],
xlabel='Time lag',
ylabel='Gain',
legend=chans,
linestyle='--',
linewidth=1,
ax=ax,
title=title)
ax.set_prop_cycle(None)
weight_x = np.arange(-1 * wcN, 0)
w_h = ax.plot(weight_x, wcc)
ax.plot(weight_x, np.array([0, 0]), linewidth=0.5, color='gray')
ax.set_xlim((-1 * wcN, len(times)))
strf_l = ['Weighted FIR {}'.format(n + 1) for n in range(wcN)]
fir_l = ['Raw FIR {}'.format(n + 1) for n in range(wcN)]
plt.legend(strf_h + fir_h, strf_l + fir_l, loc=1, fontsize='x-small')
ax.set_xticks(
np.hstack((np.arange(-1 * wcN, 0), np.arange(0,
len(times) + 1, 2))))
ax.set_xticklabels(
np.hstack((np.arange(1, wcN + 1), np.arange(0,
len(times) + 1, 2))))