def compute_metric(self, num_channels_to_compare, max_spikes_per_cluster, **kwargs):
params_dict = update_all_param_dicts_with_kwargs(kwargs)
save_property_or_features = params_dict['save_property_or_features']
seed = params_dict['seed']
l_ratios_all = metrics.calculate_pc_metrics(
spike_clusters=self._metric_data._spike_clusters_pca,
total_units=self._metric_data._total_units,
pc_features=self._metric_data._pc_features,
pc_feature_ind=self._metric_data._pc_feature_ind,
num_channels_to_compare=num_channels_to_compare,
max_spikes_for_cluster=max_spikes_per_cluster,
channel_locations=self._metric_data._channel_locations,
spikes_for_nn=None,
n_neighbors=None,
metric_names=["l_ratio"],
seed=seed,
spike_cluster_subset=self._metric_data._unit_indices,
verbose=self._metric_data.verbose,
)[1]
l_ratios_list = []
for i in self._metric_data._unit_indices:
l_ratios_list.append(l_ratios_all[i])
l_ratios = np.asarray(l_ratios_list)
if save_property_or_features:
self.save_property_or_features(self._metric_data._sorting, l_ratios, self._metric_name)
return l_ratios
def compute_metric(self, num_channels_to_compare, max_spikes_per_cluster, **kwargs):
params_dict = update_all_param_dicts_with_kwargs(kwargs)
save_property_or_features = params_dict['save_property_or_features']
seed = params_dict['seed']
isolation_distances_epochs = []
for epoch in self._metric_data._epochs:
in_epoch = self._metric_data.get_in_epoch_bool_mask(epoch, self._metric_data._spike_times_pca)
isolation_distances_all = metrics.calculate_pc_metrics(
spike_clusters=self._metric_data._spike_clusters_pca[in_epoch],
total_units=self._metric_data._total_units,
pc_features=self._metric_data._pc_features[in_epoch, :, :],
pc_feature_ind=self._metric_data._pc_feature_ind,
num_channels_to_compare=num_channels_to_compare,
max_spikes_for_cluster=max_spikes_per_cluster,
spikes_for_nn=None,
n_neighbors=None,
metric_names=["isolation_distance"],
seed=seed,
verbose=self._metric_data.verbose,
)[0]
isolation_distances_list = []
for i in self._metric_data._unit_indices:
isolation_distances_list.append(isolation_distances_all[i])
isolation_distances = np.asarray(isolation_distances_list)
isolation_distances_epochs.append(isolation_distances)
if save_property_or_features:
self.save_property_or_features(self._metric_data._sorting, isolation_distances_epochs, self._metric_name)
return isolation_distances_epochs
def compute_metric(self, num_channels_to_compare, max_spikes_per_cluster,
seed, save_as_property):
isolation_distances_epochs = []
for epoch in self._metric_data._epochs:
in_epoch = np.logical_and(
self._metric_data._spike_times_pca > epoch[1],
self._metric_data._spike_times_pca < epoch[2],
)
isolation_distances_all = metrics.calculate_pc_metrics(
spike_clusters=self._metric_data._spike_clusters_pca[in_epoch],
total_units=self._metric_data._total_units,
pc_features=self._metric_data._pc_features[in_epoch, :, :],
pc_feature_ind=self._metric_data._pc_feature_ind,
num_channels_to_compare=num_channels_to_compare,
max_spikes_for_cluster=max_spikes_per_cluster,
spikes_for_nn=None,
n_neighbors=None,
metric_names=["isolation_distance"],
seed=seed,
verbose=self._metric_data.verbose,
)[0]
isolation_distances_list = []
for i in self._metric_data._unit_indices:
isolation_distances_list.append(isolation_distances_all[i])
isolation_distances = np.asarray(isolation_distances_list)
isolation_distances_epochs.append(isolation_distances)
if save_as_property:
self.save_as_property(self._metric_data._sorting,
isolation_distances_epochs,
self._metric_name)
return isolation_distances_epochs
def compute_metric(self, num_channels_to_compare, max_spikes_per_cluster,
max_spikes_for_nn, n_neighbors, **kwargs):
params_dict = update_all_param_dicts_with_kwargs(kwargs)
save_property_or_features = params_dict['save_property_or_features']
seed = params_dict['seed']
nn_hit_rates_epochs = []
nn_miss_rates_epochs = []
for epoch in self._metric_data._epochs:
in_epoch = self._metric_data.get_in_epoch_bool_mask(
epoch, self._metric_data._spike_times_pca)
spikes_in_epoch = np.sum(in_epoch)
spikes_for_nn = np.min([spikes_in_epoch, max_spikes_for_nn])
nn_hit_rates_all, nn_miss_rates_all = metrics.calculate_pc_metrics(
spike_clusters=self._metric_data._spike_clusters_pca[in_epoch],
total_units=self._metric_data._total_units,
pc_features=self._metric_data._pc_features[in_epoch, :, :],
pc_feature_ind=self._metric_data._pc_feature_ind,
num_channels_to_compare=num_channels_to_compare,
max_spikes_for_cluster=max_spikes_per_cluster,
spikes_for_nn=spikes_for_nn,
n_neighbors=n_neighbors,
metric_names=["nearest_neighbor"],
seed=seed,
verbose=self._metric_data.verbose,
)[3:5]
nn_hit_rates_list = []
nn_miss_rates_list = []
for i in self._metric_data._unit_indices:
nn_hit_rates_list.append(nn_hit_rates_all[i])
nn_miss_rates_list.append(nn_miss_rates_all[i])
nn_hit_rates = np.asarray(nn_hit_rates_list)
nn_miss_rates = np.asarray(nn_miss_rates_list)
nn_hit_rates_epochs.append(nn_hit_rates)
nn_miss_rates_epochs.append(nn_miss_rates)
if save_property_or_features:
self.save_property_or_features(self._metric_data._sorting,
nn_hit_rates_epochs,
metric_name="nn_hit_rate")
self.save_property_or_features(self._metric_data._sorting,
nn_miss_rates_epochs,
metric_name="nn_miss_rate")
return list(zip(nn_hit_rates_epochs, nn_miss_rates_epochs))
def compute_metric(self, num_channels_to_compare, max_spikes_per_cluster,
max_spikes_for_nn, n_neighbors, seed, save_as_property):
nn_hit_rates_epochs = []
nn_miss_rates_epochs = []
for epoch in self._metric_data._epochs:
in_epoch = np.logical_and(
self._metric_data._spike_times_pca > epoch[1],
self._metric_data._spike_times_pca < epoch[2])
spikes_in_epoch = np.sum(in_epoch)
spikes_for_nn = np.min([spikes_in_epoch, max_spikes_for_nn])
nn_hit_rates_all, nn_miss_rates_all = metrics.calculate_pc_metrics(
spike_clusters=self._metric_data._spike_clusters_pca[in_epoch],
total_units=self._metric_data._total_units,
pc_features=self._metric_data._pc_features[in_epoch, :, :],
pc_feature_ind=self._metric_data._pc_feature_ind,
num_channels_to_compare=num_channels_to_compare,
max_spikes_for_cluster=max_spikes_per_cluster,
spikes_for_nn=spikes_for_nn,
n_neighbors=n_neighbors,
metric_names=["nearest_neighbor"],
seed=seed,
verbose=self._metric_data.verbose,
)[3:5]
nn_hit_rates_list = []
nn_miss_rates_list = []
for i in self._metric_data._unit_indices:
nn_hit_rates_list.append(nn_hit_rates_all[i])
nn_miss_rates_list.append(nn_miss_rates_all[i])
nn_hit_rates = np.asarray(nn_hit_rates_list)
nn_miss_rates = np.asarray(nn_miss_rates_list)
nn_hit_rates_epochs.append(nn_hit_rates)
nn_miss_rates_epochs.append(nn_miss_rates)
if save_as_property:
self.save_as_property(self._metric_data._sorting,
nn_hit_rates_epochs,
metric_name="nn_hit_rate")
self.save_as_property(self._metric_data._sorting,
nn_miss_rates_epochs,
metric_name="nn_miss_rate")
return list(zip(nn_hit_rates_epochs, nn_miss_rates_epochs))
def compute_metric(self, num_channels_to_compare, max_spikes_per_cluster,
max_spikes_for_nn, n_neighbors, **kwargs):
params_dict = update_all_param_dicts_with_kwargs(kwargs)
save_property_or_features = params_dict['save_property_or_features']
seed = params_dict['seed']
total_spikes = self._metric_data._spike_clusters_pca.shape[0]
spikes_for_nn = np.min([total_spikes, max_spikes_for_nn])
nn_hit_rates_all, nn_miss_rates_all = metrics.calculate_pc_metrics(
spike_clusters=self._metric_data._spike_clusters_pca,
total_units=self._metric_data._total_units,
pc_features=self._metric_data._pc_features,
pc_feature_ind=self._metric_data._pc_feature_ind,
num_channels_to_compare=num_channels_to_compare,
max_spikes_for_cluster=max_spikes_per_cluster,
spikes_for_nn=spikes_for_nn,
n_neighbors=n_neighbors,
channel_locations=self._metric_data._channel_locations,
metric_names=["nearest_neighbor"],
seed=seed,
spike_cluster_subset=self._metric_data._unit_indices,
verbose=self._metric_data.verbose,
)[3:5]
nn_hit_rates_list = []
nn_miss_rates_list = []
for i in self._metric_data._unit_indices:
nn_hit_rates_list.append(nn_hit_rates_all[i])
nn_miss_rates_list.append(nn_miss_rates_all[i])
nn_hit_rates = np.asarray(nn_hit_rates_list)
nn_miss_rates = np.asarray(nn_miss_rates_list)
if save_property_or_features:
self.save_property_or_features(self._metric_data._sorting,
nn_hit_rates,
metric_name="nn_hit_rate")
self.save_property_or_features(self._metric_data._sorting,
nn_miss_rates,
metric_name="nn_miss_rate")
return [nn_hit_rates_list, nn_miss_rates]