/
ecp.py
290 lines (230 loc) · 11.6 KB
/
ecp.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
# Copyright 2017. Allen Institute. All rights reserved
#
# Redistribution and use in source and binary forms, with or without modification, are permitted provided that the
# following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following
# disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following
# disclaimer in the documentation and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote
# products derived from this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
# INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#
import os
import h5py
import math
import pandas as pd
from neuron import h
import numpy as np
from bmtk.simulator.bionet.modules.sim_module import SimulatorMod
from bmtk.utils.sonata.utils import add_hdf5_magic, add_hdf5_version
pc = h.ParallelContext()
MPI_RANK = int(pc.id())
N_HOSTS = int(pc.nhost())
class EcpMod(SimulatorMod):
def __init__(self, tmp_dir, file_name, electrode_positions, contributions_dir=None, cells=None, variable_name='v',
electrode_channels=None, cell_bounds=None):
self._ecp_output = file_name if os.path.isabs(file_name) else os.path.join(tmp_dir, file_name)
self._positions_file = electrode_positions
self._tmp_outputdir = tmp_dir
if contributions_dir is not None:
self._save_individ_cells = True
self._contributions_dir = contributions_dir if os.path.isabs(contributions_dir) else os.path.join(tmp_dir, contributions_dir)
else:
self._save_individ_cells = False
self._contributions_dir = None
if cell_bounds is None:
self._cells = cells
else:
self._cells = list(np.arange(cell_bounds[0],cell_bounds[1]+1))
self._rel = None
self._fih1 = None
self._rel_nsites = 0
self._block_size = 0
self._saved_gids = {}
self._nsteps = 0
self._tstep = 0 # accumlative time step
# self._rel_time = 0 #
self._block_step = 0 # time step within the given block of time
self._tstep_start_block = 0
self._data_block = None
self._cell_var_files = {}
self._tmp_ecp_file = self._get_tmp_fname(MPI_RANK)
self._tmp_ecp_handle = None
# self._tmp_ecp_dataset = None
self._local_gids = []
def _get_tmp_fname(self, rank):
return os.path.join(self._tmp_outputdir, 'tmp_{}_ecp.h5'.format(MPI_RANK))
def _create_ecp_file(self, sim):
dt = sim.dt
tstop = sim.tstop
self._nsteps = int(round(tstop/dt))
# create file to temporary store ecp data on each rank
self._tmp_ecp_handle = h5py.File(self._tmp_ecp_file, 'a')
self._tmp_ecp_handle.create_dataset('/ecp/data', (self._nsteps, self._rel_nsites), maxshape=(None, self._rel_nsites),
chunks=True)
# only the primary node will need to save the final ecp
if MPI_RANK == 0:
with h5py.File(self._ecp_output, 'w') as f5:
add_hdf5_magic(f5)
add_hdf5_version(f5)
f5.create_dataset('/ecp/data', (self._nsteps, self._rel_nsites), maxshape=(None, self._rel_nsites),
chunks=True)
f5['/ecp/data'].attrs['units'] = 'mV'
#f5.attrs['dt'] = dt
#f5.attrs['tstart'] = 0.0
#f5.attrs['tstop'] = tstop
f5.create_dataset('/ecp/time', (3,), data=(0.0, sim.tstop, sim.dt))
f5['/ecp/time'].attrs['units'] = 'ms'
# Save channels. Current we record from all channels, may want to be more selective in the future.
f5.create_dataset('/ecp/channel_id', data=np.arange(self._rel.nsites))
pc.barrier()
def _create_cell_file(self, gid):
if not self._save_individ_cells:
return
file_name = os.path.join(self._contributions_dir, '{}.h5'.format(int(gid)))
file_h5 = h5py.File(file_name, 'a')
self._cell_var_files[gid] = file_h5
file_h5.create_dataset('/ecp/data', (self._nsteps, self._rel_nsites), maxshape=(None, self._rel_nsites), chunks=True)
# self._cell_var_files[gid] = file_h5['ecp']
def _calculate_ecp(self, sim):
self._rel = RecXElectrode(self._positions_file)
for gid in self._local_gids:
cell = sim.net.get_cell_gid(gid)
#cell = sim.net.get_local_cell(gid)
# cell = sim.net.cells[gid]
self._rel.calc_transfer_resistance(gid, cell.get_seg_coords())
self._rel_nsites = self._rel.nsites
sim.h.cvode.use_fast_imem(1) # make i_membrane_ a range variable
def set_pointers():
for gid, cell in sim.net.get_local_cells().items():
#for gid, cell in sim.net.local_cells.items():
# for gid, cell in sim.net.cells.items():
cell.set_im_ptr()
self._fih1 = sim.h.FInitializeHandler(0, set_pointers)
def _save_block(self, interval):
"""Add """
itstart, itend = interval
self._tmp_ecp_handle['/ecp/data'][itstart:itend, :] += self._data_block[0:(itend - itstart), :]
self._tmp_ecp_handle.flush()
self._data_block[:] = 0.0
def _save_ecp(self, sim):
"""Save ECP from each rank to disk into a single file"""
block_size = sim.nsteps_block
nblocks, remain = divmod(self._nsteps, block_size)
ivals = [i*block_size for i in range(nblocks+1)]
if remain != 0:
ivals.append(self._nsteps)
for rank in range(N_HOSTS): # iterate over the ranks
if rank == MPI_RANK: # wait until finished with a particular rank
with h5py.File(self._ecp_output, 'a') as ecp_f5:
for i in range(len(ivals)-1):
ecp_f5['/ecp/data'][ivals[i]:ivals[i+1], :] += self._tmp_ecp_handle['/ecp/data'][ivals[i]:ivals[i+1], :]
pc.barrier()
def _save_cell_vars(self, interval):
itstart, itend = interval
for gid, data in self._saved_gids.items():
h5_file = self._cell_var_files[gid]
h5_file['/ecp/data'][itstart:itend, :] = data[0:(itend-itstart), :]
h5_file.flush()
data[:] = 0.0
def _delete_tmp_files(self):
if os.path.exists(self._tmp_ecp_file):
self._tmp_ecp_handle.close()
os.remove(self._tmp_ecp_file)
def initialize(self, sim):
if self._contributions_dir and (not os.path.exists(self._contributions_dir)) and MPI_RANK == 0:
os.makedirs(self._contributions_dir)
pc.barrier()
self._block_size = sim.nsteps_block
# Get list of gids being recorded
selected_gids = set(sim.net.get_node_set(self._cells).gids())
self._local_gids = list(set(sim.biophysical_gids) & selected_gids)
self._calculate_ecp(sim)
self._create_ecp_file(sim)
# ecp data
self._data_block = np.zeros((self._block_size, self._rel_nsites))
# create list of all cells whose ecp values will be saved separetly
self._saved_gids = {gid: np.empty((self._block_size, self._rel_nsites))
for gid in self._local_gids}
for gid in self._saved_gids.keys():
self._create_cell_file(gid)
pc.barrier()
def step(self, sim, tstep):
for gid in self._local_gids: # compute ecp only from the biophysical cells
cell = sim.net.get_cell_gid(gid)
im = cell.get_im()
tr = self._rel.get_transfer_resistance(gid)
ecp = np.dot(tr, im)
if gid in self._saved_gids.keys():
# save individual contribution
self._saved_gids[gid][self._block_step, :] = ecp
# add to total ecp contribution
self._data_block[self._block_step, :] += ecp
self._block_step += 1
def block(self, sim, block_interval):
self._save_block(block_interval)
if self._save_individ_cells:
self._save_cell_vars(block_interval)
self._block_step = 0
self._tstep_start_block = self._tstep
def finalize(self, sim):
if self._block_step > 0:
# just in case the simulation doesn't end on a block step
self.block(sim, (sim.n_steps - self._block_step, sim.n_steps))
self._save_ecp(sim)
self._delete_tmp_files()
pc.barrier()
class RecXElectrode(object):
"""Extracellular electrode
"""
def __init__(self, positions):
"""Create an array"""
# self.conf = conf
electrode_file = positions # self.conf["recXelectrode"]["positions"]
# convert coordinates to ndarray, The first index is xyz and the second is the channel number
el_df = pd.read_csv(electrode_file, sep=' ')
self.pos = el_df[['x_pos', 'y_pos', 'z_pos']].T.values
#self.pos = el_df.as_matrix(columns=['x_pos', 'y_pos', 'z_pos']).T
self.nsites = self.pos.shape[1]
# self.conf['run']['nsites'] = self.nsites # add to the config
self.transfer_resistances = {} # V_e = transfer_resistance*Im
def drift(self):
# will include function to model electrode drift
pass
def get_transfer_resistance(self, gid):
return self.transfer_resistances[gid]
def calc_transfer_resistance(self, gid, seg_coords):
"""Precompute mapping from segment to electrode locations"""
sigma = 0.3 # mS/mm
r05 = (seg_coords['p0'] + seg_coords['p1']) / 2
dl = seg_coords['p1'] - seg_coords['p0']
nseg = r05.shape[1]
tr = np.zeros((self.nsites, nseg))
for j in range(self.nsites): # calculate mapping for each site on the electrode
rel = np.expand_dims(self.pos[:, j], axis=1) # coordinates of a j-th site on the electrode
rel_05 = rel - r05 # distance between electrode and segment centers
# compute dot product column-wise, the resulting array has as many columns as original
r2 = np.einsum('ij,ij->j', rel_05, rel_05)
# compute dot product column-wise, the resulting array has as many columns as original
rlldl = np.einsum('ij,ij->j', rel_05, dl)
dlmag = np.linalg.norm(dl, axis=0) # length of each segment
rll = abs(rlldl / dlmag) # component of r parallel to the segment axis it must be always positive
rT2 = r2 - rll ** 2 # square of perpendicular component
up = rll + dlmag / 2
low = rll - dlmag / 2
num = up + np.sqrt(up ** 2 + rT2)
den = low + np.sqrt(low ** 2 + rT2)
tr[j, :] = np.log(num / den) / dlmag # units of (um) use with im_ (total seg current)
tr *= 1 / (4 * math.pi * sigma)
self.transfer_resistances[gid] = tr