def _init_data(self):
self.cache_dir = self.config_dir
self.n_samples_waveforms = 31
self.n_samples_t = 20000
self.n_channels = 11
self.n_clusters = 4
self.n_spikes_per_cluster = 50
n_spikes_total = self.n_clusters * self.n_spikes_per_cluster
n_features_per_channel = 4
self.n_channels = self.n_channels
self.n_spikes = n_spikes_total
self.sample_rate = 20000.
self.duration = self.n_samples_t / float(self.sample_rate)
self.spike_times = np.arange(0, self.duration, 100. / self.sample_rate)
self.spike_clusters = np.repeat(np.arange(self.n_clusters),
self.n_spikes_per_cluster)
assert len(self.spike_times) == len(self.spike_clusters)
self.cluster_ids = np.unique(self.spike_clusters)
self.channel_positions = staggered_positions(self.n_channels)
sc = self.spike_clusters
self.spikes_per_cluster = lambda c: _spikes_in_clusters(sc, [c])
self.spike_count = lambda c: len(self.spikes_per_cluster(c))
self.n_features_per_channel = n_features_per_channel
self.cluster_groups = {c: None for c in range(self.n_clusters)}
self.all_traces = artificial_traces(self.n_samples_t, self.n_channels)
self.all_masks = artificial_masks(n_spikes_total, self.n_channels)
self.all_waveforms = artificial_waveforms(n_spikes_total,
self.n_samples_waveforms,
self.n_channels)
self.all_features = artificial_features(n_spikes_total,
self.n_channels,
self.n_features_per_channel)
def _init_data(self):
self.cache_dir = self.config_dir
self.n_samples_waveforms = 31
self.n_samples_t = 20000
self.n_channels = 11
self.n_clusters = 4
self.n_spikes_per_cluster = 200
n_spikes_total = self.n_clusters * self.n_spikes_per_cluster
n_features_per_channel = 4
self.n_channels = self.n_channels
self.n_spikes = n_spikes_total
self.sample_rate = 20000.
self.duration = self.n_samples_t / float(self.sample_rate)
self.spike_times = np.arange(
0, self.duration,
5000. / (self.sample_rate * self.n_spikes_per_cluster))
self.spike_clusters = np.repeat(np.arange(self.n_clusters),
self.n_spikes_per_cluster)
assert len(self.spike_times) == len(self.spike_clusters)
self.cluster_ids = np.unique(self.spike_clusters)
self.channel_positions = staggered_positions(self.n_channels)
self.channel_order = np.arange(self.n_channels)
sc = self.spike_clusters
self.spikes_per_cluster = lambda c: _spikes_in_clusters(sc, [c])
self.spike_count = lambda c: len(self.spikes_per_cluster(c))
self.n_features_per_channel = n_features_per_channel
self.cluster_groups = {c: None for c in range(self.n_clusters)}
self.all_traces = artificial_traces(self.n_samples_t, self.n_channels)
self.all_masks = artificial_masks(n_spikes_total, self.n_channels)
self.all_waveforms = artificial_waveforms(n_spikes_total,
self.n_samples_waveforms,
self.n_channels)
self.all_features = artificial_features(n_spikes_total,
self.n_channels,
self.n_features_per_channel)
def test_feature_view(qtbot, tempdir, n_channels):
nc = n_channels
ns = 500
features = artificial_features(ns, nc, 4)
spike_clusters = artificial_spike_clusters(ns, 4)
spike_times = np.linspace(0., 1., ns)
spc = _spikes_per_cluster(spike_clusters)
def get_spike_ids(cluster_id):
return (spc[cluster_id] if cluster_id is not None else np.arange(ns))
def get_features(cluster_id=None,
channel_ids=None,
spike_ids=None,
load_all=None):
if load_all:
spike_ids = spc[cluster_id]
else:
spike_ids = get_spike_ids(cluster_id)
return Bunch(
data=features[spike_ids],
spike_ids=spike_ids,
masks=np.random.rand(ns, nc),
channel_ids=(channel_ids
if channel_ids is not None else np.arange(nc)[::-1]),
)
def get_time(cluster_id=None, load_all=None):
return Bunch(
data=spike_times[get_spike_ids(cluster_id)],
lim=(0., 1.),
)
v = FeatureView(
features=get_features,
attributes={'time': get_time},
)
v.set_state(GUIState(scaling=None))
gui = GUI(config_dir=tempdir)
gui.show()
v.attach(gui)
qtbot.addWidget(gui)
v.on_select([])
v.on_select([0])
v.on_select([0, 2, 3])
v.on_select([0, 2])
gui.emit('select', [0, 2])
qtbot.wait(10)
v.increase()
v.decrease()
v.on_channel_click(channel_id=3, button=1, key=2)
v.clear_channels()
v.toggle_automatic_channel_selection()
# Split without selection.
spike_ids = v.on_request_split()
assert len(spike_ids) == 0
# Draw a lasso.
def _click(x, y):
qtbot.mouseClick(v.native,
Qt.LeftButton,
pos=QPoint(x, y),
modifier=Qt.ControlModifier)
_click(10, 10)
_click(10, 100)
_click(100, 100)
_click(100, 10)
# Split lassoed points.
spike_ids = v.on_request_split()
assert len(spike_ids) > 0
# qtbot.stop()
gui.close()
def test_feature_view(qtbot, tempdir, n_channels):
nc = n_channels
ns = 500
features = artificial_features(ns, nc, 4)
spike_clusters = artificial_spike_clusters(ns, 4)
spike_times = np.linspace(0., 1., ns)
spc = _spikes_per_cluster(spike_clusters)
def get_spike_ids(cluster_id):
return (spc[cluster_id] if cluster_id is not None else np.arange(ns))
def get_features(cluster_id=None, channel_ids=None, spike_ids=None,
load_all=None):
if load_all:
spike_ids = spc[cluster_id]
else:
spike_ids = get_spike_ids(cluster_id)
return Bunch(data=features[spike_ids],
spike_ids=spike_ids,
masks=np.random.rand(ns, nc),
channel_ids=(channel_ids
if channel_ids is not None
else np.arange(nc)[::-1]),
)
def get_time(cluster_id=None, load_all=None):
return Bunch(data=spike_times[get_spike_ids(cluster_id)],
lim=(0., 1.),
)
v = FeatureView(features=get_features,
attributes={'time': get_time},
)
v.set_state(GUIState(scaling=None))
gui = GUI(config_dir=tempdir)
gui.show()
v.attach(gui)
qtbot.addWidget(gui)
v.on_select([])
v.on_select([0])
v.on_select([0, 2, 3])
v.on_select([0, 2])
gui.emit('select', [0, 2])
qtbot.wait(10)
v.increase()
v.decrease()
v.on_channel_click(channel_id=3, button=1, key=2)
v.clear_channels()
v.toggle_automatic_channel_selection()
# Split without selection.
spike_ids = v.on_request_split()
assert len(spike_ids) == 0
# Draw a lasso.
def _click(x, y):
qtbot.mouseClick(v.native, Qt.LeftButton, pos=QPoint(x, y),
modifier=Qt.ControlModifier)
_click(10, 10)
_click(10, 100)
_click(100, 100)
_click(100, 10)
# Split lassoed points.
spike_ids = v.on_request_split()
assert len(spike_ids) > 0
# qtbot.stop()
gui.close()
def features(n_spikes, n_channels, n_features_per_channel):
yield artificial_features(n_spikes, n_channels, n_features_per_channel)
def features(n_spikes, n_channels, n_features_per_channel):
yield artificial_features(n_spikes, n_channels, n_features_per_channel)
