DARTsort 0.3.5__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- dartsort/__init__.py +68 -0
- dartsort/cli.py +105 -0
- dartsort/clustering/__init__.py +3 -0
- dartsort/clustering/cluster_util.py +392 -0
- dartsort/clustering/clustering.py +765 -0
- dartsort/clustering/clustering_features.py +202 -0
- dartsort/clustering/density.py +1299 -0
- dartsort/clustering/forward_backward.py +219 -0
- dartsort/clustering/gmm/mixture.py +5790 -0
- dartsort/clustering/gmm/stable_features.py +993 -0
- dartsort/clustering/kmeans.py +448 -0
- dartsort/clustering/merge.py +501 -0
- dartsort/clustering/modes.py +214 -0
- dartsort/clustering/refine_util.py +191 -0
- dartsort/clustering/relocate.py +109 -0
- dartsort/clustering/split.py +1285 -0
- dartsort/config.py +288 -0
- dartsort/detect/__init__.py +2 -0
- dartsort/detect/detect.py +449 -0
- dartsort/detect/detect_filters.py +37 -0
- dartsort/evaluate/__init__.py +1 -0
- dartsort/evaluate/analysis.py +594 -0
- dartsort/evaluate/comparison.py +534 -0
- dartsort/evaluate/hybrid_util.py +599 -0
- dartsort/evaluate/noise_recording_tools.py +184 -0
- dartsort/evaluate/sim_template_tools.py +842 -0
- dartsort/evaluate/simkit.py +840 -0
- dartsort/evaluate/simlib.py +270 -0
- dartsort/localize/__init__.py +0 -0
- dartsort/localize/localize_torch.py +291 -0
- dartsort/localize/localize_util.py +313 -0
- dartsort/main.py +657 -0
- dartsort/peel/__init__.py +6 -0
- dartsort/peel/grab.py +208 -0
- dartsort/peel/matching.py +628 -0
- dartsort/peel/matching_util/__init__.py +3 -0
- dartsort/peel/matching_util/compressed_upsampled.py +718 -0
- dartsort/peel/matching_util/drifty.py +791 -0
- dartsort/peel/matching_util/matching_base.py +616 -0
- dartsort/peel/matching_util/pairwise.py +359 -0
- dartsort/peel/matching_util/pairwise_util.py +1317 -0
- dartsort/peel/peel_base.py +1000 -0
- dartsort/peel/running_template.py +1498 -0
- dartsort/peel/subtract.py +1123 -0
- dartsort/peel/threshold.py +471 -0
- dartsort/peel/universal.py +191 -0
- dartsort/pretrained/default_temporal_kernel.npy +0 -0
- dartsort/pretrained/detect_np1.pt +0 -0
- dartsort/pretrained/detect_np2.pt +0 -0
- dartsort/pretrained/glasbey1024.npz +0 -0
- dartsort/pretrained/offset_trained_denoiser.pt +0 -0
- dartsort/pretrained/offset_trained_denoiser_added_temp.pt +0 -0
- dartsort/pretrained/offset_trained_denoiser_added_temp_boost.pt +0 -0
- dartsort/pretrained/original_single_chan_denoiser.pt +0 -0
- dartsort/pretrained/single_chan_denoiser.pt +0 -0
- dartsort/templates/__init__.py +11 -0
- dartsort/templates/get_templates.py +771 -0
- dartsort/templates/postprocess_util.py +555 -0
- dartsort/templates/realignment.py +466 -0
- dartsort/templates/superres_util.py +204 -0
- dartsort/templates/template_util.py +518 -0
- dartsort/templates/templates.py +528 -0
- dartsort/transform/__init__.py +2 -0
- dartsort/transform/_multichan_denoiser_kit.py +650 -0
- dartsort/transform/all_transformers.py +58 -0
- dartsort/transform/amortized_localization.py +503 -0
- dartsort/transform/amplitudes.py +197 -0
- dartsort/transform/decollider.py +635 -0
- dartsort/transform/enforce_decrease.py +158 -0
- dartsort/transform/localize.py +86 -0
- dartsort/transform/matching_denoiser.py +44 -0
- dartsort/transform/pipeline.py +391 -0
- dartsort/transform/single_channel_denoiser.py +153 -0
- dartsort/transform/single_channel_templates.py +98 -0
- dartsort/transform/supervised_denoiser.py +486 -0
- dartsort/transform/temporal_pca.py +413 -0
- dartsort/transform/transform_base.py +255 -0
- dartsort/util/__init__.py +0 -0
- dartsort/util/cli_util.py +179 -0
- dartsort/util/data_util.py +1128 -0
- dartsort/util/drift_util.py +1054 -0
- dartsort/util/internal_config.py +1074 -0
- dartsort/util/interpolation_util.py +1012 -0
- dartsort/util/job_util.py +19 -0
- dartsort/util/logging_util.py +91 -0
- dartsort/util/main_util.py +235 -0
- dartsort/util/more_operators.py +269 -0
- dartsort/util/multiprocessing_util.py +211 -0
- dartsort/util/nn_util.py +461 -0
- dartsort/util/noise_util.py +1461 -0
- dartsort/util/peel_util.py +223 -0
- dartsort/util/py_util.py +213 -0
- dartsort/util/registration_util.py +128 -0
- dartsort/util/residual_util.py +30 -0
- dartsort/util/sparse_util.py +81 -0
- dartsort/util/spikeio.py +227 -0
- dartsort/util/spiketorch.py +1198 -0
- dartsort/util/testing_util/__init__.py +0 -0
- dartsort/util/testing_util/matching_debug_util.py +517 -0
- dartsort/util/testing_util/mixture_testing_util.py +291 -0
- dartsort/util/torch_optimization_util.py +156 -0
- dartsort/util/torch_util.py +75 -0
- dartsort/util/universal_util.py +269 -0
- dartsort/util/waveform_util.py +809 -0
- dartsort/vis/__init__.py +19 -0
- dartsort/vis/analysis_plots.py +392 -0
- dartsort/vis/colors.py +36 -0
- dartsort/vis/gt.py +392 -0
- dartsort/vis/layout.py +149 -0
- dartsort/vis/metrics_util.py +23 -0
- dartsort/vis/mixture.py +1489 -0
- dartsort/vis/over_time.py +514 -0
- dartsort/vis/recanim.py +429 -0
- dartsort/vis/scatterplots.py +779 -0
- dartsort/vis/sorting.py +163 -0
- dartsort/vis/unit.py +917 -0
- dartsort/vis/unit_comparison.py +1017 -0
- dartsort/vis/versus.py +300 -0
- dartsort/vis/vismain.py +537 -0
- dartsort/vis/waveforms.py +357 -0
- dartsort-0.3.5.dist-info/METADATA +93 -0
- dartsort-0.3.5.dist-info/RECORD +125 -0
- dartsort-0.3.5.dist-info/WHEEL +5 -0
- dartsort-0.3.5.dist-info/entry_points.txt +2 -0
- dartsort-0.3.5.dist-info/top_level.txt +1 -0
|
@@ -0,0 +1,313 @@
|
|
|
1
|
+
"""Helper functions for localizing things other than torch Tensors"""
|
|
2
|
+
|
|
3
|
+
import threading
|
|
4
|
+
|
|
5
|
+
import h5py
|
|
6
|
+
import numpy as np
|
|
7
|
+
import torch
|
|
8
|
+
from tqdm.auto import tqdm
|
|
9
|
+
|
|
10
|
+
from ..util.multiprocessing_util import get_pool
|
|
11
|
+
from ..util.spiketorch import ptp
|
|
12
|
+
from .localize_torch import localize_amplitude_vectors, vmap_point_source_find_alpha
|
|
13
|
+
|
|
14
|
+
|
|
15
|
+
def localize_waveforms(
|
|
16
|
+
waveforms,
|
|
17
|
+
geom,
|
|
18
|
+
main_channels=None,
|
|
19
|
+
channel_index=None,
|
|
20
|
+
radius=None,
|
|
21
|
+
n_channels_subset=None,
|
|
22
|
+
):
|
|
23
|
+
C, dim = geom.shape
|
|
24
|
+
amp_vecs = ptp(waveforms, 1)
|
|
25
|
+
if main_channels is None:
|
|
26
|
+
main_channels = amp_vecs.argmax(1)
|
|
27
|
+
|
|
28
|
+
return localize_amplitude_vectors(
|
|
29
|
+
amp_vecs,
|
|
30
|
+
geom,
|
|
31
|
+
main_channels,
|
|
32
|
+
channel_index=channel_index,
|
|
33
|
+
radius=radius,
|
|
34
|
+
n_channels_subset=n_channels_subset,
|
|
35
|
+
)
|
|
36
|
+
|
|
37
|
+
|
|
38
|
+
def localize_hdf5(
|
|
39
|
+
hdf5_filename,
|
|
40
|
+
radius=None,
|
|
41
|
+
n_channels_subset=None,
|
|
42
|
+
output_dataset_name="point_source_localizations",
|
|
43
|
+
main_channels_dataset_name="channels",
|
|
44
|
+
amplitude_vectors_dataset_name="denoised_peak_amplitude_vectors",
|
|
45
|
+
channel_index_dataset_name="channel_index",
|
|
46
|
+
geometry_dataset_name="geom",
|
|
47
|
+
spikes_per_batch=10 * 1024,
|
|
48
|
+
overwrite=False,
|
|
49
|
+
show_progress=True,
|
|
50
|
+
device=None,
|
|
51
|
+
localization_model="pointsource",
|
|
52
|
+
logbarrier=True,
|
|
53
|
+
n_jobs=0,
|
|
54
|
+
):
|
|
55
|
+
"""Run localization on a HDF5 file with stored amplitude vectors
|
|
56
|
+
|
|
57
|
+
When this is run after a peeling pipeline which includes an AmplitudeVector
|
|
58
|
+
feature, that's equivalent to having had a PointSourceLocalization feature.
|
|
59
|
+
But, it's faster to do this after the fact.
|
|
60
|
+
"""
|
|
61
|
+
if device is None:
|
|
62
|
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
|
63
|
+
device = torch.device(device)
|
|
64
|
+
|
|
65
|
+
done, do_delete, next_batch_start = check_resume_or_overwrite(
|
|
66
|
+
hdf5_filename,
|
|
67
|
+
output_dataset_name,
|
|
68
|
+
main_channels_dataset_name,
|
|
69
|
+
overwrite=overwrite,
|
|
70
|
+
)
|
|
71
|
+
if done:
|
|
72
|
+
return
|
|
73
|
+
|
|
74
|
+
try:
|
|
75
|
+
n_jobs, Executor, context, queue = get_pool(
|
|
76
|
+
n_jobs,
|
|
77
|
+
with_rank_queue=True,
|
|
78
|
+
rank_queue_empty=True,
|
|
79
|
+
cls="ProcessPoolExecutor",
|
|
80
|
+
)
|
|
81
|
+
with Executor(
|
|
82
|
+
max_workers=n_jobs,
|
|
83
|
+
mp_context=context,
|
|
84
|
+
initializer=_h5_localize_init,
|
|
85
|
+
initargs=(
|
|
86
|
+
queue,
|
|
87
|
+
hdf5_filename,
|
|
88
|
+
main_channels_dataset_name,
|
|
89
|
+
amplitude_vectors_dataset_name,
|
|
90
|
+
channel_index_dataset_name,
|
|
91
|
+
geometry_dataset_name,
|
|
92
|
+
radius,
|
|
93
|
+
n_channels_subset,
|
|
94
|
+
localization_model,
|
|
95
|
+
device,
|
|
96
|
+
spikes_per_batch,
|
|
97
|
+
logbarrier,
|
|
98
|
+
),
|
|
99
|
+
) as pool:
|
|
100
|
+
with h5py.File(hdf5_filename, "r+", locking=False) as h5:
|
|
101
|
+
n_spikes = h5[main_channels_dataset_name].shape[0]
|
|
102
|
+
if do_delete:
|
|
103
|
+
del h5[output_dataset_name]
|
|
104
|
+
localizations_dataset = h5.require_dataset(
|
|
105
|
+
output_dataset_name,
|
|
106
|
+
shape=(n_spikes, 4),
|
|
107
|
+
dtype=np.float64,
|
|
108
|
+
fillvalue=np.nan,
|
|
109
|
+
exact=True,
|
|
110
|
+
)
|
|
111
|
+
|
|
112
|
+
# workers are blocked waiting for their ranks, which
|
|
113
|
+
# allows us to put the h5 in swmr before they open it
|
|
114
|
+
if not h5.swmr_mode:
|
|
115
|
+
h5.swmr_mode = True
|
|
116
|
+
for rank in range(n_jobs):
|
|
117
|
+
queue.put(rank)
|
|
118
|
+
|
|
119
|
+
batches = range(next_batch_start, n_spikes, spikes_per_batch)
|
|
120
|
+
results = pool.map(_h5_localize_job, batches)
|
|
121
|
+
if show_progress:
|
|
122
|
+
results = tqdm(results, total=len(batches), desc="Localization")
|
|
123
|
+
for start_ix, end_ix, xyza_batch in results:
|
|
124
|
+
localizations_dataset[start_ix:end_ix] = xyza_batch
|
|
125
|
+
pool.shutdown()
|
|
126
|
+
finally:
|
|
127
|
+
if hasattr(_loc_context, "ctx"):
|
|
128
|
+
del _loc_context.ctx
|
|
129
|
+
|
|
130
|
+
|
|
131
|
+
def check_resume_or_overwrite(
|
|
132
|
+
hdf5_filename,
|
|
133
|
+
output_dataset_name,
|
|
134
|
+
main_channels_dataset_name,
|
|
135
|
+
overwrite=False,
|
|
136
|
+
):
|
|
137
|
+
done = False
|
|
138
|
+
do_delete = False
|
|
139
|
+
next_batch_start = 0
|
|
140
|
+
with h5py.File(hdf5_filename, "r", locking=False) as h5:
|
|
141
|
+
if output_dataset_name in h5:
|
|
142
|
+
n_spikes = h5[main_channels_dataset_name].shape[0]
|
|
143
|
+
shape = h5[output_dataset_name].shape
|
|
144
|
+
if overwrite and shape != (n_spikes, 4):
|
|
145
|
+
do_delete = True
|
|
146
|
+
elif shape != (n_spikes, 4):
|
|
147
|
+
raise ValueError(
|
|
148
|
+
f"The {output_dataset_name} dataset in {hdf5_filename} "
|
|
149
|
+
f"has unexpected shape {shape}, where we expected {(n_spikes, 4)}."
|
|
150
|
+
)
|
|
151
|
+
else:
|
|
152
|
+
# else, we are resuming
|
|
153
|
+
nan_ix = np.flatnonzero(np.isnan(h5[output_dataset_name][:, 0]))
|
|
154
|
+
if nan_ix.size:
|
|
155
|
+
next_batch_start = nan_ix[0]
|
|
156
|
+
else:
|
|
157
|
+
done = True
|
|
158
|
+
return done, do_delete, next_batch_start
|
|
159
|
+
|
|
160
|
+
|
|
161
|
+
class H5LocalizationContext:
|
|
162
|
+
def __init__(
|
|
163
|
+
self,
|
|
164
|
+
hdf5_filename,
|
|
165
|
+
main_channels_dataset_name,
|
|
166
|
+
amplitude_vectors_dataset_name,
|
|
167
|
+
channel_index_dataset_name,
|
|
168
|
+
geometry_dataset_name,
|
|
169
|
+
device,
|
|
170
|
+
spikes_per_batch,
|
|
171
|
+
radius,
|
|
172
|
+
n_channels_subset,
|
|
173
|
+
localization_model,
|
|
174
|
+
logbarrier,
|
|
175
|
+
):
|
|
176
|
+
h5 = h5py.File(hdf5_filename, "r", swmr=True, locking=False)
|
|
177
|
+
channels = h5[main_channels_dataset_name][:]
|
|
178
|
+
amp_vecs = h5[amplitude_vectors_dataset_name]
|
|
179
|
+
channel_index = h5[channel_index_dataset_name][:]
|
|
180
|
+
geom = h5[geometry_dataset_name][:]
|
|
181
|
+
|
|
182
|
+
assert geom.shape == (channel_index.shape[0], 2)
|
|
183
|
+
assert amp_vecs.shape == (*channels.shape, channel_index.shape[1])
|
|
184
|
+
|
|
185
|
+
self.geom = torch.tensor(geom, device=device)
|
|
186
|
+
self.channel_index = torch.tensor(channel_index, device=device)
|
|
187
|
+
self.channels = torch.tensor(channels, device=device)
|
|
188
|
+
self.amp_vecs = amp_vecs
|
|
189
|
+
self.device = device
|
|
190
|
+
self.n_spikes = channels.shape[0]
|
|
191
|
+
self.spikes_per_batch = spikes_per_batch
|
|
192
|
+
self.radius = radius
|
|
193
|
+
self.n_channels_subset = n_channels_subset
|
|
194
|
+
self.localization_model = localization_model
|
|
195
|
+
self.logbarrier = logbarrier
|
|
196
|
+
|
|
197
|
+
|
|
198
|
+
_loc_context = threading.local()
|
|
199
|
+
|
|
200
|
+
|
|
201
|
+
def _h5_localize_init(
|
|
202
|
+
rank_queue,
|
|
203
|
+
hdf5_filename,
|
|
204
|
+
main_channels_dataset_name,
|
|
205
|
+
amplitude_vectors_dataset_name,
|
|
206
|
+
channel_index_dataset_name,
|
|
207
|
+
geometry_dataset_name,
|
|
208
|
+
radius,
|
|
209
|
+
n_channels_subset,
|
|
210
|
+
localization_model,
|
|
211
|
+
device,
|
|
212
|
+
spikes_per_batch,
|
|
213
|
+
logbarrier,
|
|
214
|
+
):
|
|
215
|
+
my_rank = rank_queue.get()
|
|
216
|
+
if device is None:
|
|
217
|
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
|
218
|
+
device = torch.device(device)
|
|
219
|
+
if device.type == "cuda" and device.index is None:
|
|
220
|
+
if torch.cuda.device_count() > 1:
|
|
221
|
+
device = torch.device("cuda", index=my_rank % torch.cuda.device_count())
|
|
222
|
+
|
|
223
|
+
_loc_context.ctx = H5LocalizationContext(
|
|
224
|
+
hdf5_filename,
|
|
225
|
+
main_channels_dataset_name,
|
|
226
|
+
amplitude_vectors_dataset_name,
|
|
227
|
+
channel_index_dataset_name,
|
|
228
|
+
geometry_dataset_name,
|
|
229
|
+
device,
|
|
230
|
+
spikes_per_batch,
|
|
231
|
+
radius,
|
|
232
|
+
n_channels_subset,
|
|
233
|
+
localization_model,
|
|
234
|
+
logbarrier,
|
|
235
|
+
)
|
|
236
|
+
|
|
237
|
+
|
|
238
|
+
def _h5_localize_job(start_ix):
|
|
239
|
+
p = _loc_context.ctx
|
|
240
|
+
|
|
241
|
+
end_ix = min(p.n_spikes, start_ix + p.spikes_per_batch)
|
|
242
|
+
channels_batch = p.channels[start_ix:end_ix]
|
|
243
|
+
amp_vecs_batch = torch.tensor(p.amp_vecs[start_ix:end_ix], device=p.device)
|
|
244
|
+
|
|
245
|
+
with torch.enable_grad():
|
|
246
|
+
locs = localize_amplitude_vectors(
|
|
247
|
+
amp_vecs_batch,
|
|
248
|
+
p.geom,
|
|
249
|
+
channels_batch,
|
|
250
|
+
channel_index=p.channel_index,
|
|
251
|
+
radius=p.radius,
|
|
252
|
+
n_channels_subset=p.n_channels_subset,
|
|
253
|
+
model=p.localization_model,
|
|
254
|
+
logbarrier=p.logbarrier,
|
|
255
|
+
)
|
|
256
|
+
xyza_batch = np.c_[
|
|
257
|
+
locs["x"].cpu().numpy(),
|
|
258
|
+
locs["y"].cpu().numpy(),
|
|
259
|
+
locs["z_abs"].cpu().numpy(),
|
|
260
|
+
locs["alpha"].cpu().numpy(),
|
|
261
|
+
]
|
|
262
|
+
return start_ix, end_ix, xyza_batch
|
|
263
|
+
|
|
264
|
+
|
|
265
|
+
def point_source_mse(locs, amp_vecs, channels, channel_index, geom):
|
|
266
|
+
is_torch = torch.is_tensor(locs)
|
|
267
|
+
locs = torch.asarray(locs)
|
|
268
|
+
amp_vecs = torch.asarray(amp_vecs)
|
|
269
|
+
channels = torch.asarray(channels)
|
|
270
|
+
channel_index = torch.asarray(channel_index)
|
|
271
|
+
pgeom = np.pad(geom, [(0, 1), (0, 0)], constant_values=np.nan)
|
|
272
|
+
geom = torch.asarray(geom)
|
|
273
|
+
pgeom = torch.asarray(pgeom)
|
|
274
|
+
|
|
275
|
+
neighborhoods = channel_index[channels]
|
|
276
|
+
invalid = neighborhoods == len(geom)
|
|
277
|
+
dxz = pgeom[neighborhoods]
|
|
278
|
+
|
|
279
|
+
if locs.shape[1] == 3:
|
|
280
|
+
channel_mask = torch.logical_not(invalid).to(torch.float)
|
|
281
|
+
alpha = vmap_point_source_find_alpha(
|
|
282
|
+
torch.asarray(amp_vecs).nan_to_num(),
|
|
283
|
+
channel_mask,
|
|
284
|
+
*torch.asarray(locs).T,
|
|
285
|
+
torch.asarray(dxz).nan_to_num(),
|
|
286
|
+
)
|
|
287
|
+
alpha = alpha.numpy()
|
|
288
|
+
else:
|
|
289
|
+
alpha = locs[:, 3]
|
|
290
|
+
locs = locs[:, :3]
|
|
291
|
+
|
|
292
|
+
x, y, z = locs.T
|
|
293
|
+
|
|
294
|
+
dxz[:, :, 0] -= x[:, None]
|
|
295
|
+
dxz[:, :, 1] -= z[:, None]
|
|
296
|
+
dxz.square_()
|
|
297
|
+
pred = dxz.sum(2)
|
|
298
|
+
pred += y[:, None] ** 2
|
|
299
|
+
pred.sqrt_()
|
|
300
|
+
pred[invalid] = 1.0
|
|
301
|
+
pred.reciprocal_()
|
|
302
|
+
pred *= alpha[:, None]
|
|
303
|
+
|
|
304
|
+
mse = amp_vecs - pred
|
|
305
|
+
mse[invalid] = 0.0
|
|
306
|
+
mse.square_()
|
|
307
|
+
nnz = channel_index.shape[1] - invalid.sum(1)
|
|
308
|
+
mse = mse.sum(1) / nnz
|
|
309
|
+
|
|
310
|
+
if not is_torch:
|
|
311
|
+
mse = mse.numpy()
|
|
312
|
+
|
|
313
|
+
return mse
|