#This is the Umat value
UMAT = u.build_u_matrix(sm, distance=1, row_normalized=False)
#Here you have Umatrix plus its render
UMAT = u.show(sm, distance2=1, row_normalized=False, show_data=True, contooor=True, blob=False)
### Try to plot a report
# dataSpikes.samples
# dataSpikes.timestamps
dataSpikes.clusters = sm.cluster()[sm.project_data(allWaves)]
figure()
plot_cluster_waves(allWaves, dataSpikes.clusters)
cr = spikesorting.ClusterReportFromData(dataSpikes, outputDir='/home/nick/Desktop', filename='somcluster_shared_{}{}{}.png'.format(animalName, ephysFn,'Tetrode{}'.format(tetrode) ))
codebookVecs = sm.codebook.matrix
from sklearn.manifold import TSNE
import timeit
start_time = timeit.default_timer()
model = TSNE(n_components=2, method='barnes_hut', verbose=20, n_iter=1000)
Y = model.fit_transform(codebookVecs)
elapsed = timeit.default_timer() - start_time
print 'ELAPSED TIME: {} mins'.format(elapsed/60)
X_pc = sklearn.decomposition.RandomizedPCA(
n_components=50).fit_transform(allWaves)
elapsed = timeit.default_timer() - start_time
print 'ELAPSED TIME: {} mins'.format(elapsed / 60)
wavesToUse = random.randint(len(X_pc), size=10000)
pcWavesToUse = X_pc[wavesToUse, :]
model = sklearn.mixture.GMM(n_components=12)
start_time = timeit.default_timer()
model.fit(pcWavesToUse)
elapsed = timeit.default_timer() - start_time
print 'ELAPSED TIME: {} mins'.format(elapsed / 60)
start_time = timeit.default_timer()
clusters = model.predict(X_pc)
elapsed = timeit.default_timer() - start_time
print 'ELAPSED TIME: {} mins'.format(elapsed / 60)
GAIN = 5000.0
SAMPLING_RATE = 30000.0
dataSpikes.samples = ((dataSpikes.samples - 32768.0) / GAIN) * 1000.0
dataSpikes.timestamps = dataSpikes.timestamps / SAMPLING_RATE
dataSpikes.clusters = clusters
spikesorting.ClusterReportFromData(dataSpikes,
outputDir='/home/nick/Desktop',
filename='test50PC_GMM.png')
from jaratoolbox import loadopenephys
reload(loadopenephys)
from pylab import *
N_CHANNELS = 4
SAMPLES_PER_SPIKE = 40
dataDir = os.path.join(settings.EPHYS_PATH,
'%s/%s/' % (animalName, ephysSession))
tetrodeFile = os.path.join(dataDir, 'Tetrode%d.spikes' % tetrode)
dataTT = loadopenephys.DataSpikes(tetrodeFile)
dataTT.timestamps = dataTT.timestamps / 0.03 # in microsec
dataTT.samples = dataTT.samples.astype(float) - 2**15
dataTT.set_clusters('/tmp/TT2.clu.1')
crep = spikesorting.ClusterReportFromData(dataTT)
'''
dataTT.samples = dataTT.samples.reshape((-1,N_CHANNELS,SAMPLES_PER_SPIKE),order='C')
fetArray = spikesorting.calculate_features(dataTT.samples,['peak','valley'])
spikesorting.write_fet_file('/tmp/TT2.fet.1',fetArray)
'''
'''
plot(dataTT.samples[:10,:].T,'.-')
draw()
show()
'''
'''
~/tmp/klustakwik/KK2/KlustaKwik TT6 1 -Subset 1e5 -MinClusters 6 -MaxClusters 12 -MaxPossibleClusters 12 -UseFeatures 11111111
def show_report(self):
self.find_cluster_each_spike()
self.dataTT.set_clusters(self.clusterEachSpike)
spikesorting.ClusterReportFromData(self.dataTT,
nrows=self.nClusters + 1)
n_iter=1000,
metric='precomputed')
Y = model.fit_transform(X_dist)
plot(Y[:, 0], Y[:, 1], '.')
GAIN = 5000.0
SAMPLING_RATE = 30000.0
dataSpikes.samples = ((dataSpikes.samples - 32768.0) / GAIN) * 1000.0
dataSpikes.timestamps = dataSpikes.timestamps / SAMPLING_RATE
from jaratoolbox import spikesorting
spikesorting.ClusterReportFromData(dataSpikes,
outputDir='/home/nick/Desktop',
filename='testKTSNEcluster.png')
# for i in range(21):
# sys.stdout.write('\r')
# # the exact output you're looking for:
# sys.stdout.write("[%-20s] %d%%" % ('='*i, 5*i))
# sys.stdout.flush()
# sleep(0.25)
# progress = (9/100.)
# sys.stdout.write("[%-20s] %d%%" % ('='*np.floor(progress*20), np.floor(progress*100)))
animalname = 'adap020'
ephysloc = '/home/nick/data/ephys/'
ephyspath = os.path.join(ephysloc, animalname)