def generate_prediction(self, model, **kwargs):
isi_var = self.get_prediction(model)
if isi_var is None:
if kwargs:
self.params.update(kwargs)
if 'variation_measure' not in self.params:
self.params.update(variation_measure='lv')
spiketrains = model.produce_spiketrains(**self.params)
isi_list = [isi(st) for st in spiketrains]
if self.params['variation_measure'] == 'lv':
isi_var = []
for intervals in isi_list:
if intervals.size > 2:
isi_var.append(lv(intervals))
elif self.params['variation_measure'] == 'cv':
isi_var = []
for intervals in isi_list:
if intervals.size > 2:
isi_var.append(cv(intervals))
elif self.params['variation_measure'] == 'isi':
isi_var = [
float(item) for sublist in isi_list for item in sublist
]
else:
raise ValueError('Variation measure not known.')
self.set_prediction(model, isi_var)
return isi_var
def test_2short_spike_train(self):
seq = [1]
with self.assertWarns(UserWarning):
"""
Catches UserWarning: Input size is too small. Please provide
an input with more than 1 entry.
"""
self.assertTrue(math.isnan(es.lv(seq, with_nan=True)))
def test_lv_with_list(self):
seq = self.test_seq
assert_array_almost_equal(es.lv(seq), self.target, decimal=9)
def test_lv_with_plain_array(self):
seq = np.array(self.test_seq)
assert_array_almost_equal(es.lv(seq), self.target, decimal=9)
def test_lv_with_quantities(self):
seq = pq.Quantity(self.test_seq, units='ms')
assert_array_almost_equal(es.lv(seq), self.target, decimal=9)
def generate_prediction(self, model=None):
prediction = lv(model.get_membrane_potential())
return prediction
return prediction
def test_lv_with_plain_array(self):
seq = np.array(self.test_seq)
assert_array_almost_equal(es.lv(seq), self.target, decimal=9)
def test_lv_with_list(self):
seq = self.test_seq
assert_array_almost_equal(es.lv(seq), self.target, decimal=9)
def test_lv_with_quantities(self):
seq = pq.Quantity(self.test_seq, units='ms')
assert_array_almost_equal(es.lv(seq), self.target, decimal=9)
def generate_prediction(self, model = None):
prediction = lv(model.get_membrane_potential())
return prediction
return prediction
def main(p,file,save_path):
pre = 10*pq.ms
post = 10*pq.ms
fid = PIO(file)
blk = fid.read_block()
FR,ISI,contact_trains = get_contact_sliced_trains(blk,pre=pre,post=post)
binsize = 2*pq.ms
for unit in blk.channel_indexes[-1].units:
root = blk.annotations['ratnum'] + blk.annotations['whisker'] + 'c{}'.format(unit.name[-1])
trains = contact_trains[unit.name]
all_isi = np.array([])
CV_array = np.array([])
LV_array = np.array([])
for interval in ISI[unit.name]:
all_isi = np.concatenate([all_isi,interval])
if np.all(np.isfinite(interval)):
CV_array = np.concatenate([CV_array,[cv(interval)]])
LV_array = np.concatenate([LV_array,[lv(interval)]])
all_isi = all_isi * interval.units
CV_array = CV_array
CV = np.mean(CV_array)
LV = np.mean(LV_array)
## calculate data for PSTH
b,durations = get_binary_trains(contact_trains[unit.name])
b_times = np.where(b)[1] * pq.ms#interval.units
b_times-=pre
PSTH,t_edges = np.histogram(b_times,bins=np.arange(-np.array(pre),np.max(durations)+np.array(post),float(binsize)))
plt.bar(t_edges[:-1],
PSTH.astype('f8')/len(durations)/binsize*1000,
width=float(binsize),
align='edge',
alpha=0.8
)
ax = plt.gca()
thresh = 500 * pq.ms
ax.set_xlim(-15, thresh.__int__())
ax.set_xlabel('Time after contact (ms)')
ax.set_ylabel('Spikes per second')
ax.set_title('PSTH for: {}'.format(root))
plt.savefig(os.path.join(save_path,root+'_PSTH.svg'))
plt.close('all')
# ============================================
# PLOT ISIs
plt.figure()
thresh = 100 * pq.ms
if len(all_isi[np.logical_and(np.isfinite(all_isi), all_isi < thresh)])==0:
return
ax = sns.distplot(all_isi[np.logical_and(np.isfinite(all_isi), all_isi < thresh)],
bins=np.arange(0,100,1),
kde_kws={'color':'k','lw':3,'alpha':0.5,'label':'KDE'})
ax.set_xlabel('ISI '+all_isi.dimensionality.latex)
ax.set_ylabel('Percentage of all ISIs')
a_inset = plt.axes([.55, .5, .2, .2], facecolor='w')
a_inset.grid(color='k',linestyle=':',alpha=0.4)
a_inset.axvline(CV,color='k',lw=0.5)
a_inset.set_title('CV = {:0.2f}\nLV = {:0.2f}'.format(CV,LV))
a_inset.set_xlabel('CV')
a_inset.set_ylabel('# of Contacts')
sns.distplot(CV_array,color='g',kde=False)
ax.set_title('ISI distribution for {}'.format(root))
plt.savefig(os.path.join(save_path, root + '_ISI.svg'))
plt.close('all')
def test_2short_spike_train(self):
seq = [1]
self.assertTrue(math.isnan(es.lv(seq, with_nan=True)))
def test_2short_spike_train(self):
seq = [1]
self.assertTrue(math.isnan(es.lv(seq, with_nan=True)))