def test_cell_class_init():
with raises(AssertionError):
CellClass(asdf=1)
with raises(AssertionError):
CellClass(db.Pair.n_ex_test_spikes < 10)
with raises(AssertionError):
CellClass([])
with raises(AssertionError):
CellClass(name=1)
def test_cell_class_contains():
cell = db.Cell()
morpho = db.Morphology()
loc = db.CorticalCellLocation()
cell.morphology = morpho
cell.cortical_location = loc
cls = CellClass(cre_type=('sst', 'pvalb'))
cell.cre_type = 'sst'
assert cell in cls
cell.cre_type = 'sim1'
assert cell not in cls
cls = CellClass(cre_type='sim1', dendrite_type='spiny')
morpho.dendrite_type = 'spiny'
assert cell in cls
morpho.dendrite_type = 'aspiny'
assert cell not in cls
cls = CellClass(cre_type='sim1', dendrite_type='spiny', cortical_layer='2/3')
morpho.dendrite_type = 'spiny'
loc.cortical_layer = '2/3'
assert cell in cls
loc.cortical_layer = '2'
assert cell not in cls
cls = CellClass(cre_type='sim1', dendrite_type='spiny', cortical_layer='2/3', cell_class='ex')
cell.cell_class = 'ex'
loc.cortical_layer = '2/3'
assert cell in cls
cell.cell_class = 'mixed'
assert cell not in cls
cls = CellClass(input_resistance=0)
# no intrinsic added to this cell
assert cell not in cls
def ei_hist_plot(ax, metric, bin_edges, db, pair_query_args):
ei_classes = {'ex': CellClass(cell_class='ex'), 'in': CellClass(cell_class='in')}
pairs_has_metric, metric_name, units, scale, _, _, log_scale, _, _ = get_metric_data(metric, db, ei_classes, ei_classes, pair_query_args=pair_query_args)
ex_pairs = pairs_has_metric[pairs_has_metric['pre_class']=='ex']
in_pairs = pairs_has_metric[pairs_has_metric['pre_class']=='in']
if 'amp' in metric:
ax[0].hist(ex_pairs[metric]*scale, bins=bin_edges, color=(0.8, 0.8, 0.8), label='All Excitatory Synapses')
ax[1].hist(in_pairs[metric]*scale, bins=bin_edges, color=(0.8, 0.8, 0.8), label='All Inhibitory Synapses')
else:
ax[0].hist(pairs_has_metric[metric]*scale, bins=bin_edges, color=(0.8, 0.8, 0.8), label='All Synapses')
ax[1].hist(pairs_has_metric[metric]*scale, bins=bin_edges, color=(0.8, 0.8, 0.8), label='All Synapses')
ee_pairs = ex_pairs[ex_pairs['post_class']=='ex']
ei_pairs = ex_pairs[ex_pairs['post_class']=='in']
ax[0].hist(ee_pairs[metric]*scale, bins=bin_edges, color='red', alpha=0.6, label='E->E Synapses')
ax[0].hist(ei_pairs[metric]*scale, bins=bin_edges, color='pink', alpha=0.8, label='E->I Synapses')
ax[0].legend(frameon=False)
ii_pairs = in_pairs[in_pairs['post_class']=='in']
ie_pairs = in_pairs[in_pairs['post_class']=='ex']
ax[1].hist(ii_pairs[metric]*scale, bins=bin_edges, color='blue', alpha=0.4, label='I->I Synapses')
ax[1].hist(ie_pairs[metric]*scale, bins=bin_edges, color='purple', alpha=0.4, label='I->E Synapses')
ax[1].legend(frameon=False)
ax[0].spines['right'].set_visible(False)
ax[0].spines['top'].set_visible(False)
ax[1].spines['right'].set_visible(False)
ax[1].spines['top'].set_visible(False)
ax[1].set_xlabel('%s (%s)' % (metric_name, units))
ax[1].set_ylabel('Number of Synapses', fontsize=12)
#KS test
excitatory = stats.ks_2samp(ee_pairs[metric], ei_pairs[metric])
inhibitory = stats.ks_2samp(ii_pairs[metric], ie_pairs[metric])
print('Two-sample KS test for %s' % metric)
print('Excitatory: p = %0.3e' % excitatory[1])
print('Inhibitory: p = %0.3e' % inhibitory[1])
return ex_pairs, in_pairs
def get_pre_or_post_classes(self, key):
"""Return a list of postsynaptic cell_classes. This will be a subset of self.cell_classes."""
# if self.cell_classes is None:
# return []
ccg = copy.deepcopy(self.cell_class_groups)
classes = []
for group in self.params.children():
if group.value() is True:
if group['pre/post'] in ['both', key]:
classes.extend(ccg[group.name()])
classes = self.layer_call(classes)
classes = [CellClass(**c) for c in classes]
return classes
def get_cell_groups(self, pairs):
"""Given a list of cell pairs, return a dict indicating which cells
are members of each user selected cell class.
This internally calls cell_class.classify_cells
"""
ccg = copy.deepcopy(self.cell_class_groups)
if self.cell_groups is None:
self.cell_classes = []
for group in self.params.children()[1:]:
if group.value() is True:
self.cell_classes.extend(ccg[group.name()])
self.cell_classes = self.layer_call(self.cell_classes)
self.cell_classes = [CellClass(**c) for c in self.cell_classes]
self.cell_groups = classify_cells(self.cell_classes, pairs=pairs)
return self.cell_groups, self.cell_classes
def _make_cell_class(self, spec):
spec = spec.copy()
dnames = spec.pop('display_names')
cell_cls = CellClass(**spec, name=' '.join(dnames).strip())
cell_cls.display_names = dnames
return cell_cls
def _make_cell_class(self, spec):
spec = spec.copy()
dnames = spec.pop('display_names')
cell_cls = CellClass(**spec)
cell_cls.display_names = dnames
return cell_cls
def test_correction_model():
# test preparations
ei_classes = {
'excit': CellClass(cell_class_nonsynaptic='ex', name='ex'),
'inhib': CellClass(cell_class_nonsynaptic='in', name='in')
}
correction_metrics = {
'lateral_distance': {
'model': GaussianModel,
'init': (0.1, 100e-6),
'bounds': ((0.001, 1), (100e-6, 100e-6))
},
'pre_axon_length': {
'model': BinaryModel,
'init': (0.1, 200e-6, 0.5),
'bounds': ((0.001, 1), (200e-6, 200e-6), (0.1, 0.9))
},
'avg_pair_depth': {
'model': ErfModel,
'init': (0.1, 30e-6, 30e-6),
'bounds': ((0.01, 1), (10e-6, 200e-6), (-100e-6, 100e-6))
},
'detection_power': {
'init': (0.1, 1.0, 3.0),
'model': ErfModel,
'bounds': ((0.001, 1), (0.1, 5), (2, 5)),
'constraint': (0.6745, 4.6613)
},
}
# test finite connection cases
pairs, variables, conn = data_snippets()
corr_model = ei_correct_connectivity(ei_classes, correction_metrics, pairs)
# fixing the correction parameters from a larger pool of data
correction_parameters = [
[
np.array([0.1, 100e-6]), # lateral_distance (Gaussian)
np.array([6.939e-2, 2.000e-4,
6.003e-1]), # pre_axon_length (Binary)
np.array([6.893e-2, 4.021e-5, 2.848e-5]), # avg_pair_depth (Erf)
np.array([9.847e-2, 6.845e-1, 4.200])
], # detection_power (Erf)
[
np.array([0.1, 100e-6]),
np.array([1.227e-1, 2.000e-4, 3.634e-1]),
np.array([1.095e-1, 4.130e-5, 2.700e-5]),
np.array([1.967e-1, 4.396e-1, 4.365])
]
]
corr_model.correction_parameters = correction_parameters
assert corr_model.pmax == 0.1
fit = corr_model.fit(variables, conn, excinh=1)
assert corr_model.pmax == pytest.approx(0.74590, 1e-4)
assert fit.x == pytest.approx(0.74590, 1e-4)
assert corr_model.likelihood(variables,
conn) == pytest.approx(-1.95075, 1e-4)
assert fit.fun == pytest.approx(1.95075, 1e-4)
assert fit.cp_ci[0] == corr_model.pmax
assert fit.cp_ci[1] == pytest.approx(0.0478545, 1e-4) # lower MINOS value
assert fit.cp_ci[2] == pytest.approx(2.17707, 1e-4) # upper MINOS value
# test zero-connection cases
pairs, variables, conn = data_snippets('zero')
fit = CorrectionModel.fit(corr_model, variables, conn, excinh=1)
assert fit.cp_ci[0] == pytest.approx(0, abs=1e-6)
assert fit.cp_ci[1] == pytest.approx(0, abs=1e-6) # lower MINOS value
assert fit.cp_ci[2] == pytest.approx(1.03155, 1e-4) # upper MINOS value
# test full-cnnection cases
pairs, variables, conn = data_snippets('full')
fit = CorrectionModel.fit(corr_model, variables, conn, excinh=1)
assert fit.cp_ci[0] == pytest.approx(3, 1e-4) # maxed out
assert fit.cp_ci[1] == pytest.approx(1.87319, 1e-4) # lower MINOS value
assert fit.cp_ci[2] == pytest.approx(3, 1e-4) # upper MINOS value
return
app.processEvents()
mouse_dist_plots[0].setXRange(20e-6, 180e-6, padding=0)
mouse_dist_plots[0].setYRange(0, 0.4, padding=0)
human_dist_plots[0].setYRange(0, 0.4, padding=0)
mouse_hist_plots[0].setYRange(0, 200)
human_hist_plots[0].setYRange(0, 50)
app.processEvents()
# define CellClass, short name, and color for each class
mouse_classes = [
(CellClass(target_layer='2/3', pyramidal=True), 'L2/3', (249, 144, 92)),
(CellClass(cre_type='rorb'), 'Rorb', (100, 202, 103)),
(CellClass(cre_type='tlx3'), 'Tlx3', (81, 221, 209)),
(CellClass(cre_type='sim1'), 'Sim1', (45, 77, 247)),
(CellClass(cre_type='ntsr1'), 'Ntsr1', (153, 51, 255)),
]
human_classes = [
(CellClass(target_layer='2', pyramidal=True), 'L2', (247, 118, 118)),
(CellClass(target_layer='3', pyramidal=True), 'L3', (246, 197, 97)),
(CellClass(target_layer='4', pyramidal=True), 'L4', (100, 202, 103)),
(CellClass(target_layer='5', pyramidal=True), 'L5', (107, 155, 250)),
(CellClass(target_layer='6', pyramidal=True), 'L6', (153, 51, 255)),
]
def test_cell_class_is_excitatory():
cls = CellClass(cre_type='sst')
assert cls.is_excitatory is False
assert cls.output_synapse_type == 'in'
cls = CellClass(cre_type='sim1')
assert cls.is_excitatory is True
assert cls.output_synapse_type == 'ex'
cls = CellClass(cre_type=['sst', 'pvalb'])
assert cls.is_excitatory is False
assert cls.output_synapse_type == 'in'
cls = CellClass(cre_type=['sst', 'tlx3'])
assert cls.is_excitatory is None
assert cls.output_synapse_type is None
cls = CellClass(cre_type=['sst'], dendrite_type='spiny')
assert cls.is_excitatory is None
assert cls.output_synapse_type is None
cls = CellClass(cre_type='rorb', dendrite_type='spiny')
assert cls.is_excitatory is True
assert cls.output_synapse_type == 'ex'
cls = CellClass(cre_type='rorb', dendrite_type='sparsely spiny')
assert cls.is_excitatory is None
assert cls.output_synapse_type is None
cls = CellClass(cre_type=['sst', 'vip'], dendrite_type='sparsely spiny')
assert cls.is_excitatory is False
assert cls.output_synapse_type == 'in'
cls = CellClass(cell_class='in')
assert cls.is_excitatory is False
assert cls.output_synapse_type == 'in'
cls = CellClass(cell_class_nonsynaptic='in')
assert cls.is_excitatory is False
assert cls.output_synapse_type == 'in'
cls = CellClass(cell_class='ex')
assert cls.is_excitatory is True
assert cls.output_synapse_type == 'ex'
cls = CellClass(cell_class_nonsynaptic='ex')
assert cls.is_excitatory is True
assert cls.output_synapse_type == 'ex'
cls = CellClass(cell_class='in', cell_class_nonsynaptic='ex')
assert cls.is_excitatory is None
assert cls.output_synapse_type is None
cls = CellClass(target_layer='2/3')
assert cls.is_excitatory is None
assert cls.output_synapse_type is None
cls = CellClass(target_layer='2/3', cell_class='ex')
assert cls.is_excitatory is True
assert cls.output_synapse_type == 'ex'