def run(self):
from spikeforest import SFMdaRecordingExtractor, SFMdaSortingExtractor
print('test1', self.firings_path, self.samplerate)
sorting = SFMdaSortingExtractor(firings_file=self.firings_path)
samplerate = self.samplerate
max_samples = self.max_samples
max_dt_msec = self.max_dt_msec
bin_size_msec = self.bin_size_msec
max_dt_tp = max_dt_msec * samplerate / 1000
bin_size_tp = bin_size_msec * samplerate / 1000
autocorrelograms = []
for unit_id in sorting.get_unit_ids():
print('Unit::g {}'.format(unit_id))
(bin_counts, bin_edges) = compute_autocorrelogram(sorting.get_unit_spike_train(unit_id), max_dt_tp=max_dt_tp, bin_size_tp=bin_size_tp, max_samples=max_samples)
autocorrelograms.append(dict(
unit_id=unit_id,
bin_counts=bin_counts,
bin_edges=bin_edges
))
ret = dict(
autocorrelograms=autocorrelograms
)
with open(self.json_out, 'w') as f:
json.dump(serialize_np(ret), f)
def run(self):
rx = SFMdaRecordingExtractor(
dataset_directory=self.recording_directory,
download=True,
raw_fname=self.filtered_timeseries)
sx_true = SFMdaSortingExtractor(firings_file=self.firings_true)
sx = SFMdaSortingExtractor(firings_file=self.firings_sorted)
ssobj = create_spikesprays(rx=rx,
sx_true=sx_true,
sx_sorted=sx,
neighborhood_size=self.neighborhood_size,
num_spikes=self.num_spikes,
unit_id_true=self.unit_id_true,
unit_id_sorted=self.unit_id_sorted)
with open(self.json_out, 'w') as f:
json.dump(ssobj, f)
def run(self):
recording = SFMdaRecordingExtractor(
dataset_directory=self.recording_directory, download=True)
sorting = SFMdaSortingExtractor(firings_file=self.firings)
waveforms0 = _get_random_spike_waveforms(recording=recording,
sorting=sorting,
unit=self.unit_id)
channel_ids = recording.get_channel_ids()
avg_waveform = np.median(waveforms0, axis=2)
ret = dict(channel_ids=channel_ids,
average_waveform=avg_waveform.tolist())
with open(self.json_out, 'w') as f:
json.dump(ret, f)
def yass_example(download=True, set_id=1):
if set_id in range(1, 7):
dsdir = 'kbucket://15734439d8cf/groundtruth/visapy_mea/set{}'.format(
set_id)
IX = SFMdaRecordingExtractor(dataset_directory=dsdir,
download=download)
path1 = os.path.join(dsdir, 'firings_true.mda')
print(path1)
OX = SFMdaSortingExtractor(path1)
return (IX, OX)
else:
raise Exception(
'Invalid ID for yass_example {} is not betewen 1..6'.format(
set_id))
def initialize(self):
if self._initialized:
return
self._initialized = True
# self._recording_context.initialize()
print('******** FORESTVIEW: Initializing sorting result context')
if self._sorting_result_object['firings']:
self._sorting_extractor = SFMdaSortingExtractor(
firings_file=self._sorting_result_object['firings'])
else:
self._sorting_extractor = None
print('******** FORESTVIEW: Done initializing sorting result context')
def load_sorting_results_info(firings_path, *, recording_path, epoch_name, ntrode_name, curated=False):
if not mt.findFile(firings_path):
return None
sorting = SFMdaSortingExtractor(firings_file=firings_path)
total_num_events = 0
for unit_id in sorting.get_unit_ids():
spike_times = sorting.get_unit_spike_train(unit_id=unit_id)
total_num_events = total_num_events + len(spike_times)
return dict(
type='sorting_results',
epoch_name=epoch_name,
ntrode_name=ntrode_name,
curated=curated,
firings_path=firings_path,
recording_path=recording_path,
unit_ids=sorting.get_unit_ids(),
num_events=total_num_events
)
#!/usr/bin/env python
from spikeforest import SFMdaRecordingExtractor, SFMdaSortingExtractor
from mountaintools import client as mt
# Configure to download from the public spikeforest kachery node
mt.configDownloadFrom('spikeforest.public')
# Load an example tetrode recording with its ground truth
# You can also substitute any of the other available recordings
recdir = 'sha1dir://fb52d510d2543634e247e0d2d1d4390be9ed9e20.synth_magland/datasets_noise10_K10_C4/001_synth'
print('loading recording...')
recording = SFMdaRecordingExtractor(dataset_directory=recdir, download=True)
sorting_true = SFMdaSortingExtractor(firings_file=recdir + '/firings_true.mda')
# import a spike sorter from the spikesorters module of spikeforest
from spikeforestsorters import MountainSort4
import os
import shutil
# In place of MountainSort4 you could use any of the following:
#
# MountainSort4, SpykingCircus, KiloSort, KiloSort2, YASS
# IronClust, HerdingSpikes2, JRClust, Tridesclous, Klusta
# although the Matlab sorters require further setup.
# clear and create an empty output directory (keep things tidy)
if os.path.exists('test_outputs'):
shutil.rmtree('test_outputs')
os.makedirs('test_outputs', exist_ok=True)
from spikeforest import SFMdaRecordingExtractor, SFMdaSortingExtractor
from mountaintools import client as mt
# Configure to download from the public spikeforest kachery node
mt.configDownloadFrom('spikeforest.public')
# Load the recording with its ground truth
recdir = 'sha1dir://be6ce9f60fe1963af235862dc8197c9753b4b6f5.hybrid_janelia/drift_siprobe/rec_16c_1200s_11'
print('Loading recording...')
recording = SFMdaRecordingExtractor(dataset_directory=recdir, download=True)
sorting_true = SFMdaSortingExtractor(firings_file=recdir + '/firings_true.mda')
sorting_ms4 = SFMdaSortingExtractor(
firings_file=
'sha1://f1c6fdf52a2873d6f746e44dab6bf7ccd2937d97/f1c6fdf52a2873d6f746e44dab6bf7ccd2937d97/firings.mda'
)
# import from the spikeforest package
import spikeforest_analysis as sa
# write the ground truth firings file
SFMdaSortingExtractor.write_sorting(sorting=sorting_true,
save_path='test_outputs/firings_true.mda')
# run the comparison
print('Compare with truth...')
import time
timer = time.time()
## Old method
def sorting(self):
return SFMdaSortingExtractor(firings_file=self._obj['firings'])
def sortingTrue(self):
return SFMdaSortingExtractor(firings_file=self.directory() +
'/firings_true.mda')
from spikeforest import SFMdaRecordingExtractor, SFMdaSortingExtractor
from mountaintools import client as mt
# Configure to download from the public spikeforest kachery node
mt.configDownloadFrom('spikeforest.public')
# Load an example tetrode recording with its ground truth
# You can also substitute any of the other available recordings
recdir = 'sha1dir://fb52d510d2543634e247e0d2d1d4390be9ed9e20.synth_magland/datasets_noise10_K10_C4/001_synth'
print('loading recording...')
recording = SFMdaRecordingExtractor(dataset_directory=recdir, download=True)
sorting_true = SFMdaSortingExtractor(firings_file=recdir + '/firings_true.mda')
# clear and create an empty output directory (keep things tidy)
if os.path.exists('test_outputs'):
shutil.rmtree('test_outputs')
os.makedirs('test_outputs', exist_ok=True)
# Run spike sorting in the default singularity container
print('Spike sorting...')
MountainSort4.execute(recording_dir=recdir,
firings_out='test_outputs/ms4_firings.mda',
detect_sign=-1,
adjacency_radius=50,
_container='default')
# Load the result into a sorting extractor
sorting = SFMdaSortingExtractor(firings_file='test_outputs/ms4_firings.mda')
# import from the spikeforest package
import spikeforest_analysis as sa
# write the ground truth firings file
SFMdaSortingExtractor.write_sorting(sorting=sorting_true,
save_path='test_outputs/firings_true.mda')
# run the comparison
print('Compare with truth...')
sa.GenSortingComparisonTable.execute(
firings='test_outputs/ms4_firings.mda',
firings_true='test_outputs/firings_true.mda',
units_true=[], # use all units
json_out='test_outputs/comparison.json',
