def _get_color(clusters_data, cl):
if cl not in clusters_data:
return next_color(cl)
cluster_data = clusters_data[cl]
try:
out = cluster_data.application_data.klustaviewa.color or 1
return out
except AttributeError as e:
return next_color(cl)
def _get_color(clusters_data, cl):
if cl not in clusters_data:
return next_color(cl)
cluster_data = clusters_data[cl]
try:
out = cluster_data.application_data.klustaviewa.color or 1
return out
except AttributeError as e:
return next_color(cl)
def get_clusterview_data(
exp,
statscache=None,
channel_group=0,
clustering='main',
):
clusters_data = getattr(exp.channel_groups[channel_group].clusters,
clustering)
cluster_groups_data = getattr(
exp.channel_groups[channel_group].cluster_groups, clustering)
# Get the list of all existing clusters.
# clusters = sorted(clusters_data.keys())
spike_clusters = getattr(exp.channel_groups[channel_group].spikes.clusters,
clustering)[:]
clusters = np.unique(spike_clusters)
groups = cluster_groups_data.keys()
# cluster_groups = pd.Series([clusters_data[cl].cluster_group or 0
# for cl in clusters], index=clusters)
# Make sure there's no crash if this is called before the clusters had a chance
# to be added in the HDF5 file.
# get colors from application data:
cluster_colors = pd.Series(
[_get_color(clusters_data, cl) for cl in clusters], index=clusters)
# cluster_colors = pd.Series([
# next_color(cl)
# if cl in clusters_data else 1
# for cl in clusters], index=clusters)
cluster_groups = pd.Series(
[(clusters_data[cl].cluster_group if clusters_data[cl].cluster_group
is not None else 3) if cl in clusters_data else 3
for cl in clusters],
index=clusters)
group_colors = pd.Series([next_color(cl) for g in groups], index=groups)
group_names = pd.Series(
[cluster_groups_data[g].name or 'Group' for g in groups], index=groups)
# TODO: cache the cluster size instead of recomputing every time here
# (in experiment class?)
spike_clusters = getattr(exp.channel_groups[channel_group].spikes.clusters,
clustering)[:]
sizes = np.bincount(spike_clusters)
cluster_sizes = pd.Series(sizes[clusters], index=clusters)
data = dict(
cluster_colors=cluster_colors,
cluster_groups=cluster_groups,
group_colors=group_colors,
group_names=group_names,
cluster_sizes=cluster_sizes,
)
if statscache is not None:
data['cluster_quality'] = statscache.cluster_quality
return data
def get_clusterview_data(exp, statscache=None, channel_group=0,
clustering='main',):
clusters_data = getattr(exp.channel_groups[channel_group].clusters, clustering)
cluster_groups_data = getattr(exp.channel_groups[channel_group].cluster_groups, clustering)
# Get the list of all existing clusters.
# clusters = sorted(clusters_data.keys())
spike_clusters = getattr(exp.channel_groups[channel_group].spikes.clusters,
clustering)[:]
clusters = np.unique(spike_clusters)
groups = cluster_groups_data.keys()
# cluster_groups = pd.Series([clusters_data[cl].cluster_group or 0
# for cl in clusters], index=clusters)
# Make sure there's no crash if this is called before the clusters had a chance
# to be added in the HDF5 file.
# get colors from application data:
cluster_colors = pd.Series([_get_color(clusters_data, cl)
for cl in clusters], index=clusters)
# cluster_colors = pd.Series([
# next_color(cl)
# if cl in clusters_data else 1
# for cl in clusters], index=clusters)
cluster_groups = pd.Series([
(clusters_data[cl].cluster_group
if clusters_data[cl].cluster_group is not None else 3)
if cl in clusters_data else 3
for cl in clusters], index=clusters)
group_colors = pd.Series([next_color(cl)
for g in groups], index=groups)
group_names = pd.Series([cluster_groups_data[g].name or 'Group'
for g in groups], index=groups)
# TODO: cache the cluster size instead of recomputing every time here
# (in experiment class?)
spike_clusters = getattr(exp.channel_groups[channel_group].spikes.clusters,
clustering)[:]
sizes = np.bincount(spike_clusters)
cluster_sizes = pd.Series(sizes[clusters], index=clusters)
data = dict(
cluster_colors=cluster_colors,
cluster_groups=cluster_groups,
group_colors=group_colors,
group_names=group_names,
cluster_sizes=cluster_sizes,
)
if statscache is not None:
data['cluster_quality'] = statscache.cluster_quality
return data
def get_traceview_data(exp,
channel_group=0, clustering='main'):
if (len(exp.recordings) == 0) or exp.recordings[0].raw == None:
data = dict(
trace=None,
)
return data
rawdata = exp.recordings[0].raw
freq = exp.application_data.spikedetekt.sample_rate
clusters_data = getattr(exp.channel_groups[channel_group].clusters, clustering)
clusters = sorted(clusters_data.keys())
spikes_data = exp.channel_groups[channel_group].spikes
channels = exp.channel_groups[channel_group].channel_order
spiketimes = spikes_data.time_samples
spikeclusters = getattr(spikes_data.clusters, clustering)[:]
_, nsamples, nchannels = spikes_data.waveforms_filtered.shape
freq = exp.application_data.spikedetekt.sample_rate
cluster_colors = pd.Series([next_color(cl)
for cl in clusters], index=clusters)
fetdim = exp.application_data.spikedetekt.n_features_per_channel
s_before = exp.application_data.spikedetekt.extract_s_before
s_after = exp.application_data.spikedetekt.extract_s_after
if spikes_data.masks is not None:
spikemasks = np.zeros((spikes_data.masks.shape[0], rawdata.shape[1]))
spikemasks[:,channels] = spikes_data.masks[:, 0:fetdim*nchannels:fetdim]
cluster_colors = pandaize(cluster_colors, clusters)
data = dict(
freq=freq,
trace=rawdata,
spiketimes=spiketimes,
spikemasks=spikemasks,
spikeclusters=spikeclusters,
cluster_colors = cluster_colors,
s_before = s_before,
s_after = s_after
)
return data
def get_traceview_data(exp, channel_group=0, clustering='main'):
if (len(exp.recordings) == 0) or exp.recordings[0].raw == None:
data = dict(trace=None, )
return data
rawdata = exp.recordings[0].raw
freq = exp.application_data.spikedetekt.sample_rate
clusters_data = getattr(exp.channel_groups[channel_group].clusters,
clustering)
clusters = sorted(clusters_data.keys())
spikes_data = exp.channel_groups[channel_group].spikes
channels = exp.channel_groups[channel_group].channel_order
spiketimes = spikes_data.time_samples
spikeclusters = getattr(spikes_data.clusters, clustering)[:]
_, nsamples, nchannels = spikes_data.waveforms_filtered.shape
freq = exp.application_data.spikedetekt.sample_rate
cluster_colors = pd.Series([next_color(cl) for cl in clusters],
index=clusters)
fetdim = exp.application_data.spikedetekt.n_features_per_channel
s_before = exp.application_data.spikedetekt.extract_s_before
s_after = exp.application_data.spikedetekt.extract_s_after
if spikes_data.masks is not None:
spikemasks = np.zeros((spikes_data.masks.shape[0], rawdata.shape[1]))
spikemasks[:,
channels] = spikes_data.masks[:,
0:fetdim * nchannels:fetdim]
cluster_colors = pandaize(cluster_colors, clusters)
data = dict(freq=freq,
trace=rawdata,
spiketimes=spiketimes,
spikemasks=spikemasks,
spikeclusters=spikeclusters,
cluster_colors=cluster_colors,
s_before=s_before,
s_after=s_after)
return data
def test_colors_1():
for c in xrange(1, COLORS_COUNT):
assert next_color(c) == c + 1
assert next_color(COLORS_COUNT) == 1
def get_cluster_color(self, cluster):
try:
return next_color(cluster)
except IndexError:
return 0
def get_cluster_color(self, cluster):
try:
return next_color(cluster)
except IndexError:
return 0
