-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain_nbMatConsensus.py
208 lines (179 loc) · 10.6 KB
/
main_nbMatConsensus.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
''' This script runs the analysis comparing how changing the number of matrices for the consensus changes the variability of the harmonics'''
import seaborn as sns
import numpy as np
import scipy.io as sio
import pandas as pd
import scipy
import matplotlib.pyplot as plt
import lib.func_reading as reading
import lib.func_SDI as sdi
import lib.func_ML as ML
from lib.func_plot import plot_rois, plot_rois_pyvista
import random
from lib import fcn_groups_bin
import pandas as pd
from sklearn.metrics.pairwise import cosine_similarity
config_path = 'config.json' #'config_PEP3.json'
config_defaults = reading.check_config_file(config_path)
### Reading the data
SC = sio.loadmat('./data/Individual_Connectomes.mat')
#print(np.shape(SC['connMatrices']['SC'][0][0][1][0]))
SC = SC['connMatrices']['SC'][0][0][1][0]
roi_info_path = 'data/label/roi_info.xlsx'
roi_info = pd.read_excel(roi_info_path, sheet_name=f'SCALE 2')
cort_rois = np.where(roi_info['Structure'] == 'cort')[0]
matMetric = SC
x = np.asarray(roi_info['x-pos'])[cort_rois]
y = np.asarray(roi_info['y-pos'])[cort_rois]
z = np.asarray(roi_info['z-pos'])[cort_rois]
coordMat = np.concatenate((x[:,None],y[:,None],z[:,None]),1)
Euc = scipy.spatial.distance.squareform(scipy.spatial.distance.pdist(coordMat, metric='euclidean'))
### Generate random group based on different number of participants
total_participant = np.shape(matMetric)[2]; nROIs = np.shape(matMetric)[0]
idxs = list(range(total_participant))
#ls_bins = [2,10,20,30,40,50]
ls_bins = [10, 50]
nbPerm = 100
nbins = 41
hemii = np.ones(len(Euc))
hemii[int(len(hemii)/2):] = 2
RandCons = np.zeros((nROIs, nROIs, nbPerm, len(ls_bins)))
ShuffIdxs = np.zeros((len(idxs), nbPerm, len(ls_bins)))
for b,bi in enumerate(ls_bins):
for p in np.arange(nbPerm):
random.shuffle(idxs)
ShuffIdxs[:,p,b] = idxs
idxs_tmp = idxs[0:bi]
[G, Gc] = fcn_groups_bin.fcn_groups_bin(matMetric[:,:, idxs_tmp], Euc, hemii, nbins)
avg = np.mean(matMetric[:,:, idxs_tmp], 2)
RandCons[:,:,p,b] = Gc*avg
print('nROIs=%d, number of bins=%d, number of randomization=%d'%(np.shape(RandCons)[0], np.shape(RandCons)[3], np.shape(RandCons)[2]))
### Generate the eigenvectors
nb_eig2keep = nROIs
eigenvectors_perm = np.zeros((len(cort_rois), nb_eig2keep, len(ls_bins)*nbPerm))
eigenvalues_perm = np.zeros((nb_eig2keep, len(ls_bins)*nbPerm))
eigenvectors_perm_mat = np.zeros((len(cort_rois), nb_eig2keep, len(ls_bins), nbPerm))
eigenvalues_perm_mat = np.zeros((nb_eig2keep, len(ls_bins), nbPerm))
labels_perm = []
k = 0
for b,bi in enumerate(ls_bins):
for p in np.arange(nbPerm):
eigenvalues_perm_mat[:, b, p], eigenvectors_perm_mat[:, :, b, p],Ln_ind, An_ind = ML.cons_normalized_lap(RandCons[:,:,p,b], Euc, plot=False)
labels_perm.append('Bin%d'%(bi))
k = k+1
### rotate for each bin only
eigenvalues_perm_mat_rot = np.zeros(np.shape(eigenvalues_perm_mat))
eigenvectors_perm_mat_rot = np.zeros(np.shape(eigenvectors_perm_mat))
R_all = np.zeros(np.shape(eigenvectors_perm_mat));
scale_R = np.zeros((len(ls_bins), nbPerm)); disparity = np.copy(scale_R)
for b,bi in enumerate(ls_bins):
print(bi)
### Generalized Procrustes
# eigenvectors_perm_mat_rot[:,:,b,:], eigenvalues_perm_mat_rot[:,b,:], R_all[:,:,b,:], scale_R[b,:] = ML.rotation_procrustes(eigenvectors_perm_mat[:,:,b,:], eigenvalues_perm_mat[:,b,:], plot=True, p='bin%d'%bi)
### Orthogonal Procrustes
eigenvectors_perm_mat_rot[:,:,b,:], eigenvalues_perm_mat_rot[:,b,:], R_all[:,:,b,:], scale_R[b,:] = ML.orthogonal_rotation_procrustes(eigenvectors_perm_mat[:,:,b,:], eigenvalues_perm_mat[:,b,:], plot=True, p='bin%d'%bi)
cos_sim_rot = np.diag(cosine_similarity(eigenvectors_perm_mat_rot[:,:,b,0], eigenvectors_perm_mat_rot[:,:,b,5]))
cos_sim = np.diag(cosine_similarity(eigenvectors_perm_mat[:,:,b,0], eigenvectors_perm_mat[:,:,b,5]))
plt.figure()
plt.plot(np.abs(cos_sim))
plt.plot(cos_sim_rot)
plt.legend(['Before alignment', 'After alignment'])
plt.title('Pairwise Cosine Similarity %s'%bi)
plt.grid('on')
plt.xlabel('Eigenvector'); plt.ylabel('Cosine Similarity')
print('ok')
##for p in np.arange(nbPerm):
## eigenvectors_perm_mat_rot[:,:,:,p],eigenvalues_perm_mat_rot[:,:,p], R_all, scale_R = ML.rotation_procrustes(eigenvectors_perm_mat[:,:,:,p], eigenvalues_perm_mat[:,:,p], plot=True, p='bin%d'%p)#eigenvectors_perm = np.reshape(eigenvectors_perm_mat, ((len(cort_rois), nb_eig2keep, len(ls_bins)*nbPerm)))
eigenvectors_perm_rot = np.reshape(eigenvectors_perm_mat_rot, ((len(cort_rois), nb_eig2keep, len(ls_bins)*nbPerm)))
eigenvectors_perm = np.reshape(eigenvectors_perm_mat, ((len(cort_rois), nb_eig2keep, len(ls_bins)*nbPerm)))
X_RS_allPat = sdi.load_EEG_example()
P_ind, Q_ind, Ln_ind, An_ind = ML.cons_normalized_lap(matMetric, Euc, plot=False)
### Rotate for each bin and perm
#eigenvectors_perm = np.reshape(eigenvectors_perm_mat, ((len(cort_rois), nb_eig2keep, len(ls_bins)*nbPerm)))
#eigenvectors_perm_rot = np.zeros(np.shape(eigenvectors_perm))
#eigenvalues_perm_rot = np.zeros((nb_eig2keep, len(ls_bins)*nbPerm))
#eigenvectors_perm_rot, eigenvalues_perm_rot, R_all, scale_R, disparity = ML.rotation_procrustes(eigenvectors_perm, eigenvalues_perm, plot=True, p='bin%d'%p)
#fig, axs = plt.subplots(1,1)
#axs.scatter(eigenvectors_perm_rot[:,50,50], eigenvectors_perm[:,50,50])
#plt.show()
#eig=1
#plot_rois_pyvista(eigenvectors_perm[:,eig,50]*10, config_defaults["Parameters"]["scale"], config_defaults, vmin=-2.5, vmax=2.5, label='NotRotated%d'%eig)
#plot_rois_pyvista(eigenvectors_perm_rot[:,eig,50]*100, config_defaults["Parameters"]["scale"], config_defaults, vmin=-2.5, vmax=2.5, label='Rotated%d'%eig)
#corr_eigvec_perm_before_after_rot = np.zeros((nb_eig2keep, len(ls_bins)*nbPerm, len(ls_bins)*nbPerm))
#for eig in np.arange(nb_eig2keep):
# for k1 in np.arange(len(ls_bins)*nbPerm):
# for k2 in np.arange(len(ls_bins)*nbPerm):
# tmp = np.corrcoef(eigenvectors_perm_rot[:,eig,k1],eigenvectors_perm_rot[:,eig,k2])
# corr_eigvec_perm_before_after_rot[eig, k1,k2] = tmp[0,1]
#fig, axs = plt.subplots(1,1)
#axs.imshow(corr_eigvec_perm_before_after_rot[48,:,:]);
#plt.show()
### Generate the corresponding labels
labels_perm_mat = []
labels_perm_bin = []
for i in np.arange(len(labels_perm)):
for j in np.arange(len(labels_perm)):
labels_perm_mat.append('%s_%s'%(labels_perm[i], labels_perm[j]))
if labels_perm[i]==labels_perm[j]:
labels_perm_bin.append('%s'%(labels_perm[i]))
else:
labels_perm_bin.append('Different bins')
labels_perm_bin = np.array(labels_perm_bin)
labels_perm_mat = np.array(labels_perm_mat)
### Compute the similarity betwen all the eigenvectors (all bins and randomization)
Dist_eigvec_perm = np.zeros((len(ls_bins)*nbPerm, len(ls_bins)*nbPerm, nb_eig2keep))
Dist_eigvec_perm_rot = np.zeros((len(ls_bins)*nbPerm, len(ls_bins)*nbPerm, nb_eig2keep))
for eigvec_nb in np.arange(nb_eig2keep):
MatDist = 1 - scipy.spatial.distance.pdist(np.transpose(eigenvectors_perm[:, eigvec_nb,:]), metric='correlation')
#MatDist = scipy.spatial.distance.pdist(np.transpose(eigenvectors_perm[:, eigvec_nb,:]), metric='euclidean')
Dist_eigvec_perm[:,:,eigvec_nb] = scipy.spatial.distance.squareform(MatDist)
#MatDist_rot = scipy.spatial.distance.pdist(np.transpose(eigenvectors_perm_rot[:, eigvec_nb,:]), metric='euclidean')
MatDist_rot = 1 - scipy.spatial.distance.pdist(np.transpose(eigenvectors_perm_rot[:, eigvec_nb,:]), metric='correlation')
Dist_eigvec_perm_rot[:,:,eigvec_nb] = scipy.spatial.distance.squareform(MatDist_rot)
#### Remove the 0 values corresponding to the similarity between identical vectors
Dist_eigvec_perm_vec = np.reshape(Dist_eigvec_perm, (len(ls_bins)*nbPerm*len(ls_bins)*nbPerm, nb_eig2keep))
Dist_eigvec_perm_vec = np.abs(Dist_eigvec_perm_vec) ### Take absolute values for compensating for sign change
Dist_eigvec_perm_rot_vec = np.reshape(Dist_eigvec_perm_rot, (len(ls_bins)*nbPerm*len(ls_bins)*nbPerm, nb_eig2keep))
Dist_eigvec_perm_rot_vec = np.abs(Dist_eigvec_perm_rot_vec) ### Take absolute values for compensating for sign change
tmp2 = Dist_eigvec_perm_rot_vec[:,30]; tmp2 = tmp2[np.where(tmp2>0)]
### Remove the 0 values corresponding here to the diagonal
for i in np.arange(nb_eig2keep):
idxs_nz = np.where(Dist_eigvec_perm_vec[:,i])
tmp = Dist_eigvec_perm_vec[:,i]; tmp = tmp[idxs_nz]
tmp2 = Dist_eigvec_perm_rot_vec[:,i]; tmp2 = tmp2[idxs_nz]
if i==0:
Dist_eigvec_perm_vec_nz = np.zeros((len(tmp), nb_eig2keep))
Dist_eigvec_perm_rot_vec_nz = np.zeros((len(tmp2), nb_eig2keep))
Dist_eigvec_perm_vec_nz[:,i] = tmp
Dist_eigvec_perm_rot_vec_nz[:,i] = tmp2
labels_perm_bin = labels_perm_bin[idxs_nz]
labels_perm_mat = labels_perm_mat[idxs_nz]
bin_variability = np.zeros((len(ls_bins), nb_eig2keep, 2))
bin_variability_rot = np.zeros((len(ls_bins), nb_eig2keep, 2))
for b,bi in enumerate(ls_bins):
idxs = np.where(labels_perm_mat=='Bin%d_Bin%d'%(bi,bi))[0]
for i in np.arange(nb_eig2keep):
#print(np.median(Dist_eigvec_perm_vec_nz[idxs,i]))
bin_variability[b,i,0] = np.median(Dist_eigvec_perm_vec_nz[idxs,i])
bin_variability[b,i,1] = np.std(Dist_eigvec_perm_vec_nz[idxs,i])
#idxs_rot = np.where(labels_perm_rot_bin=='Bin%d'%bi)[0]
#for i in np.arange(nb_eig2keep):
bin_variability_rot[b,i,0] = np.median(Dist_eigvec_perm_rot_vec_nz[idxs,i])
bin_variability_rot[b,i,1] = np.std(Dist_eigvec_perm_rot_vec_nz[idxs,i])
fig, ax = plt.subplots(2,1,figsize=(15,5))
handles = []; handles_rot=[] # To store the handles for lines in the plot
for b,bi in enumerate(ls_bins):
line, = ax[0].plot(bin_variability[b,:,0]); handles.append(line);
line_rot, = ax[1].plot(bin_variability_rot[b,:,0]); handles_rot.append(line_rot);
upper_bound = bin_variability[b, :, 0] + bin_variability[b, :, 1]
lower_bound = bin_variability[b, :, 0] - bin_variability[b, :, 1]
upper_bound_rot = bin_variability_rot[b, :, 0] + bin_variability_rot[b, :, 1]
lower_bound_rot = bin_variability_rot[b, :, 0] - bin_variability_rot[b, :, 1]
ax[0].fill_between(range(nb_eig2keep), lower_bound, upper_bound, alpha=0.1)
ax[1].fill_between(range(nb_eig2keep), lower_bound_rot, upper_bound_rot, alpha=0.1)
for x in range(2):
ax[x].set_xlabel('Eigenmode'); ax[x].set_xticks(range(0, nb_eig2keep, 10))
ax[x].grid('on'); ax[x].set_ylim([0,1.02]); ax[x].set_ylabel('Correlation'); ax[x].set_title('Similarity between network harmonics', fontsize=15);
ax[0].legend(handles=handles, labels=ls_bins, loc='lower left', title='Bin');
ax[1].legend(handles=handles_rot, labels=ls_bins, loc='lower left', title='Bin');
plt.show()