/
21_supp_check_firstlevel_distributions.py
220 lines (175 loc) · 7.52 KB
/
21_supp_check_firstlevel_distributions.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
"""This script produces a table summarising first level distributions."""
import os
from copy import deepcopy
import glmtools as glm
import h5py
import matplotlib.pyplot as plt
import mne
import numpy as np
import osl
import sails
import lemon_plotting
from glm_config import cfg
from lemon_support import (get_eeg_data, lemon_make_bads_regressor,
lemon_make_blinks_regressor,
lemon_make_task_regressor)
#%% ----------------------------------------------------------
# GLM-Prep
fbase = os.path.join(cfg['lemon_processed_data'], '{subj}/{subj}_preproc_raw.fif')
st = osl.utils.Study(fbase)
fname = st.get(subj='sub-010060')[0]
runname = fname.split('/')[-1].split('.')[0]
print('processing : {0}'.format(runname))
subj_id = osl.utils.find_run_id(fname)
raw = mne.io.read_raw_fif(fname, preload=True)
picks = mne.pick_types(raw.info, eeg=True, ref_meg=False)
chlabels = np.array(raw.info['ch_names'], dtype=h5py.special_dtype(vlen=str))[picks]
fbase = os.path.join(cfg['lemon_processed_data'], 'sub-010002', 'sub-010002_preproc_raw.fif')
reference = mne.io.read_raw_fif(fbase).pick_types(eeg=True)
#%% ----------------------------------------------------------
# GLM-Prep
# Make blink regressor
blink_vect, numblinks, evoked_blink = lemon_make_blinks_regressor(raw, figpath=None)
veog = np.abs(raw.get_data(picks='ICA-VEOG')[0, :])
heog = np.abs(raw.get_data(picks='ICA-HEOG')[0, :])
# Make task regressor
task = lemon_make_task_regressor({'raw': raw})
# Make bad-segments regressor
bads = lemon_make_bads_regressor(raw)
print('Bad segs in regressor {} / {}'.format(bads.sum(), len(bads)))
# Get data
XX = get_eeg_data(raw)
print(XX.shape)
# Run GLM-Periodogram
conds = {'Eyes Open': task > 0, 'Eyes Closed': task < 0}
covs = {'Linear Trend': np.linspace(0, 1, raw.n_times)}
confs = {'Bad Segs': bads,
'V-EOG': veog, 'H-EOG': heog}
eo_val = np.round(np.sum(task == 1) / len(task), 3)
ec_val = np.round(np.sum(task == -1) / len(task), 3)
conts = [{'name': 'RestMean', 'values': {'Eyes Open': 0.5, 'Eyes Closed': 0.5}},
{'name': 'Open>Closed', 'values': {'Eyes Open': 1, 'Eyes Closed': -1}}]
fs = raw.info['sfreq']
# Reduced model - no confounds or covariates
glmspec_mag = osl.glm.glm_spectrum(XX, fmin=1, fmax=95,
fs=fs,
fit_intercept=False,
nperseg=int(fs * 2),
mode='magnitude',
contrasts=conts,
reg_categorical=conds,
reg_ztrans=covs, reg_unitmax=confs,
standardise_data=False)
glmspec_mag = osl.glm.SensorGLMSpectrum(glmspec_mag, reference.info) # Store with standard channel info
# Reduced model - no confounds or covariates
glmspec_pow = osl.glm.glm_spectrum(XX, fmin=1, fmax=95,
fs=fs,
fit_intercept=False,
nperseg=int(fs * 2),
mode='psd',
contrasts=conts,
reg_categorical=conds,
reg_ztrans=covs, reg_unitmax=confs,
standardise_data=False)
glmspec_pow = osl.glm.SensorGLMSpectrum(glmspec_pow, reference.info) # Store with standard channel info
# Reduced model - no confounds or covariates
glmspec_log2 = osl.glm.glm_spectrum(XX, fmin=1, fmax=95,
fs=fs,
fit_intercept=False,
nperseg=int(fs * 2),
mode='log_psd',
contrasts=conts,
reg_categorical=conds,
reg_ztrans=covs, reg_unitmax=confs,
standardise_data=False)
glmspec_log2 = osl.glm.SensorGLMSpectrum(glmspec_log2, reference.info) # Store with standard channel info
# Reduced model - no confounds or covariates
glmspec_log = deepcopy(glmspec_pow)
glmspec_log.data.data = np.log(glmspec_log.data.data)
glmspec_log.model = glm.fit.OLSModel(glmspec_log.design, glmspec_log.data)
#%% ----------------------------------------------------------
# Figure
plt.figure(figsize=(16, 9))
plt.subplots_adjust(wspace=0.45, hspace=0.35, left=0.04, right=0.975)
ax = plt.subplot(2, 3, 1)
glmspec_pow.plot_joint_spectrum(0, base=0.5, freqs=(3, 9 , 25), ax=ax)
ax = plt.subplot(2, 3, 2)
glmspec_mag.plot_joint_spectrum(0, base=0.5, freqs=(3, 9 , 25), ax=ax, ylabel='Magnitude')
ax = plt.subplot(2, 3, 3)
glmspec_log.plot_joint_spectrum(0, base=0.5, freqs=(3, 9 , 25), ax=ax, ylabel='log(Power)')
ax = plt.subplot(2, 3, 4)
glmspec_pow.plot_joint_spectrum(1, base=0.5, freqs=(3, 9 , 25), ax=ax, metric='tstats')
ax = plt.subplot(2, 3, 5)
glmspec_mag.plot_joint_spectrum(1, base=0.5, freqs=(3, 9 , 25), ax=ax, metric='tstats')
ax = plt.subplot(2, 3, 6)
glmspec_log.plot_joint_spectrum(1, base=0.5, freqs=(3, 9 , 25), ax=ax, metric='tstats')
fout = os.path.join(cfg['lemon_figures'], 'lemon-supp_first-level-distribution-comparison.png')
plt.savefig(fout, transparent=True, dpi=300)
fout = os.path.join(cfg['lemon_figures'], 'lemon-supp_first-level-distribution-comparison_low-res.png')
plt.savefig(fout, transparent=True, dpi=100)
eye
data_pow = deepcopy(data)
data_pow.data = data_pow.data**2
model_pow = glm.fit.OLSModel(design, data_pow)
model_mag = glm.fit.OLSModel(design, data)
data_logpow = deepcopy(data)
data_logpow.data = np.log(data_logpow.data**2)
model_logpow = glm.fit.OLSModel(design, data_logpow)
#%% ----------------------------------------------------------
# Figure
ff = 19
chan = 24
def quick_decorate(ax):
"""Decorate an axes."""
for tag in ['top', 'right']:
ax.spines[tag].set_visible(False)
ax.set_ylabel('Num Segments')
plt.figure(figsize=(12, 12))
plt.subplots_adjust(hspace=0.5)
plt.subplot(3, 3, 1)
plt.plot(data_pow.data.mean(axis=0))
lemon_plotting.decorate_spectrum(plt.gca(), ylabel='Power')
plt.title('Data Spectrum')
lemon_plotting.subpanel_label(plt.gca(), 'A')
plt.subplot(3, 3, 2)
plt.hist(data_pow.data[:, ff, chan], 64)
quick_decorate(plt.gca())
plt.xlabel('Power')
plt.title('Data Distribution\n(single channel and freq)')
plt.subplot(3, 3, 3)
resids = model_pow.get_residuals(data_pow.data)
plt.hist(resids[:, ff, chan], 64)
quick_decorate(plt.gca())
plt.xlabel('Power')
plt.title('Residual Distribution\n(single channel and freq)')
plt.subplot(3, 3, 4)
plt.plot(data.data.mean(axis=0))
lemon_plotting.decorate_spectrum(plt.gca(), ylabel='Magnitude')
lemon_plotting.subpanel_label(plt.gca(), 'B')
plt.subplot(3, 3, 5)
plt.hist(data.data[:, ff, chan], 64)
quick_decorate(plt.gca())
plt.xlabel('Magnitude')
plt.subplot(3, 3, 6)
resids = model_mag.get_residuals(data.data)
plt.hist(resids[:, ff, chan], 64)
quick_decorate(plt.gca())
plt.xlabel('Magnitude')
plt.subplot(3, 3, 7)
plt.plot(data_logpow.data.mean(axis=0))
lemon_plotting.decorate_spectrum(plt.gca(), ylabel='log(Power)')
lemon_plotting.subpanel_label(plt.gca(), 'C')
plt.subplot(3, 3, 8)
plt.hist(data_logpow.data[:, ff, chan], 64)
quick_decorate(plt.gca())
plt.xlabel('log(Power)')
plt.subplot(3, 3, 9)
resids = model_logpow.get_residuals(data_logpow.data)
plt.hist(resids[:, ff, chan], 64)
quick_decorate(plt.gca())
plt.xlabel('log(Power)')
plt.savefig('dist_check.png')
fout = os.path.join(cfg['lemon_figures'], 'lemon-supp_first-level-distribution-check.png')
plt.savefig(fout, transparent=True, dpi=300)
fout = os.path.join(cfg['lemon_figures'], 'lemon-supp_first-level-distribution-check_low-res.png')
plt.savefig(fout, transparent=True, dpi=100)