-
Notifications
You must be signed in to change notification settings - Fork 0
/
EncodEfficiency.m
245 lines (172 loc) · 9.81 KB
/
EncodEfficiency.m
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
%% This code is designed to look at the potential correlation between
% condition-indepdent variance and cortex encoding efficiency
% In this code, we look at V1 activity in the stimulus period of visual trials,
% A1 activity in the stimulus period of auditory trials, and motor cortex
% activity in the response period of all trials.
%% Only correct trials are used.
function [] = EncodEfficiency(raw_data, V1_activity, M2_activity, Group_LowVariance, Group_HighVariance, plot_index)
if isempty(plot_index)
return;
end
if isempty(V1_activity) || isempty(M2_activity)
disp('No imaging data was found! Can''t run function EncodEfficiency.');
return;
end
global mouse_name
mousename = mouse_name;
%% 1. Only getting correct trials
outcome_highV = raw_data.Rewarded(Group_HighVariance); % Group_HighVariance & Group_LowVariance don't contain any non-response trials.
outcome_lowV = raw_data.Rewarded(Group_LowVariance);
sample_highV = Group_HighVariance(outcome_highV == 1); % only using correct trials
sample_lowV = Group_LowVariance(outcome_lowV == 1);
clear outcome_highV outcome_lowV
%% 2. Applying classifier on V1, distinguishing stimulus.
if plot_index(1) == 1
sample_highV_LStim = sample_highV(raw_data.CorrectSide(sample_highV) == 1);
sample_highV_RStim = sample_highV(raw_data.CorrectSide(sample_highV) == 2);
sample_lowV_LStim = sample_lowV(raw_data.CorrectSide(sample_lowV) == 1);
sample_lowV_RStim = sample_lowV(raw_data.CorrectSide(sample_lowV) == 2);
% for balancing classifier sample numbers
sample_num = min([length(sample_highV_LStim), length(sample_highV_RStim), length(sample_lowV_LStim), length(sample_lowV_RStim)]);
% V1_activity(:,:,1:44,:) = []; % stimulus period is only from frame 45 to frame 60
% V1_activity(:,:,16:end,:) = [];
% highV group first
Accuracy_data_highV = nan(size(V1_activity,3), 10); % all frames, 30 times repeated cross validation
Accuracy_shuffled_highV = nan(size(V1_activity,3), 10);
for t = 1 : size(Accuracy_data_highV, 2)
msize = numel(sample_highV_LStim);
idx_highV_LStim = randperm(msize);
idx_highV_LStim = sample_highV_LStim(idx_highV_LStim(1:sample_num));
msize = numel(sample_highV_RStim);
idx_highV_RStim = randperm(msize);
idx_highV_RStim = sample_highV_RStim(idx_highV_RStim(1:sample_num));
idx_highV_LR = [idx_highV_LStim, idx_highV_RStim];
keys = raw_data.CorrectSide(idx_highV_LR);
V1_activity_highV = V1_activity(:,:,:,idx_highV_LR);
for frame = 1 : size(V1_activity_highV,3)
thisFrame = reshape(squeeze(V1_activity_highV(:,:,frame,:)), [], sample_num*2);
ClassifierModel = fitclinear(thisFrame, keys, 'ObservationsIn', 'columns', 'Crossval', 'on');
Loss_experiment = kfoldLoss(ClassifierModel);
ShuffleModel = fitclinear(thisFrame, keys(randperm(length(keys))), 'ObservationsIn', 'columns', 'Crossval', 'on');
Loss_control = kfoldLoss(ShuffleModel);
Accuracy_data_highV(frame, t) = (1 - Loss_experiment)*100;
Accuracy_shuffled_highV(frame, t) = (1 - Loss_control)*100;
clear ClassifierModel ShuffleModel thisFrame
end
end
% Then lowV group
Accuracy_data_lowV = nan(size(V1_activity,3), 10); % all frames, 10 times repeated cross validation
Accuracy_shuffled_lowV = nan(size(V1_activity,3), 10);
for t = 1 : size(Accuracy_data_lowV, 2)
msize = numel(sample_lowV_LStim);
idx_lowV_LStim = randperm(msize);
idx_lowV_LStim = sample_lowV_LStim(idx_lowV_LStim(1:sample_num));
msize = numel(sample_lowV_RStim);
idx_lowV_RStim = randperm(msize);
idx_lowV_RStim = sample_lowV_RStim(idx_lowV_RStim(1:sample_num));
idx_lowV_LR = [idx_lowV_LStim, idx_lowV_RStim];
keys = raw_data.CorrectSide(idx_lowV_LR);
V1_activity_lowV = V1_activity(:,:,:,idx_lowV_LR);
for frame = 1 : size(V1_activity_lowV,3)
thisFrame = reshape(squeeze(V1_activity_lowV(:,:,frame,:)), [], sample_num*2);
ClassifierModel = fitclinear(thisFrame, keys, 'ObservationsIn', 'columns', 'Crossval', 'on');
Loss_experiment = kfoldLoss(ClassifierModel);
ShuffleModel = fitclinear(thisFrame, keys(randperm(length(keys))), 'ObservationsIn', 'columns', 'Crossval', 'on');
Loss_control = kfoldLoss(ShuffleModel);
Accuracy_data_lowV(frame, t) = (1 - Loss_experiment)*100;
Accuracy_shuffled_lowV(frame, t) = (1 - Loss_control)*100;
clear ClassifierModel ShuffleModel thisFrame
end
end
figure;
curve1 = stdshade(Accuracy_data_highV',0.4,'b');
ylim([30 100]);
hold on;
curve2 = stdshade(Accuracy_shuffled_highV',0.4,'r');
curve3 = stdshade(Accuracy_data_lowV',0.4,'c');
curve4 = stdshade(Accuracy_shuffled_lowV',0.4,'m');
legend([curve1, curve2, curve3, curve4], 'highV', 'highV shuffled', 'lowV', 'lowV shuffled');
xlabel('Frames from Stimulation On');
ylabel('Accuracy(%)');
TITLE = ['Linear Classifier for predicting the stimulus based on PPC, ', mousename];
title(TITLE);
set(gca,'box','off');
set(gca,'tickdir','out');
hold off
clear curve1 curve2 curve3 curve4 TITLE
end
%% 3. Applying classifier on M2, distinguishing choice.
if plot_index(2) == 1
sample_highV_LChoice = sample_highV(raw_data.ResponseSide(sample_highV) == 1);
sample_highV_RChoice = sample_highV(raw_data.ResponseSide(sample_highV) == 2);
sample_lowV_LChoice = sample_lowV(raw_data.ResponseSide(sample_lowV) == 1);
sample_lowV_RChoice = sample_lowV(raw_data.ResponseSide(sample_lowV) == 2);
% for balancing classifier sample numbers
sample_num = min([length(sample_highV_LChoice), length(sample_highV_RChoice), length(sample_lowV_LChoice), length(sample_lowV_RChoice)]);
% M2_activity(:,:,1:60,:) = []; % response period is after frame 67
% highV group first
Accuracy_data_highV = nan(size(M2_activity,3), 10); % frames number, 30 times repeated cross validation
Accuracy_shuffled_highV = nan(size(M2_activity,3), 10);
for t = 1 : size(Accuracy_data_highV, 2)
msize = numel(sample_highV_LChoice);
idx_highV_LChoice = randperm(msize);
idx_highV_LChoice = sample_highV_LChoice(idx_highV_LChoice(1:sample_num));
msize = numel(sample_highV_RChoice);
idx_highV_RChoice = randperm(msize);
idx_highV_RChoice = sample_highV_RChoice(idx_highV_RChoice(1:sample_num));
idx_highV_LR = [idx_highV_LChoice, idx_highV_RChoice];
keys = raw_data.ResponseSide(idx_highV_LR);
M2_activity_highV = M2_activity(:,:,:,idx_highV_LR);
for frame = 1 : size(M2_activity,3)
thisFrame = reshape(squeeze(M2_activity_highV(:,:,frame,:)), [], sample_num*2);
ClassifierModel = fitclinear(thisFrame, keys, 'ObservationsIn', 'columns', 'Crossval', 'on');
Loss_experiment = kfoldLoss(ClassifierModel);
ShuffleModel = fitclinear(thisFrame, keys(randperm(length(keys))), 'ObservationsIn', 'columns', 'Crossval', 'on');
Loss_control = kfoldLoss(ShuffleModel);
Accuracy_data_highV(frame, t) = (1 - Loss_experiment)*100;
Accuracy_shuffled_highV(frame, t) = (1 - Loss_control)*100;
clear ClassifierModel ShuffleModel thisFrame
end
end
% Then lowV group
Accuracy_data_lowV = nan(size(M2_activity,3), 10); % frames number, 30 times repeated cross validation
Accuracy_shuffled_lowV = nan(size(M2_activity,3), 10);
for t = 1 : size(Accuracy_data_lowV, 2)
msize = numel(sample_lowV_LChoice);
idx_lowV_LChoice = randperm(msize);
idx_lowV_LChoice = sample_lowV_LChoice(idx_lowV_LChoice(1:sample_num));
msize = numel(sample_lowV_RChoice);
idx_lowV_RChoice = randperm(msize);
idx_lowV_RChoice = sample_lowV_RChoice(idx_lowV_RChoice(1:sample_num));
idx_lowV_LR = [idx_lowV_LChoice, idx_lowV_RChoice];
keys = raw_data.ResponseSide(idx_lowV_LR);
M2_activity_lowV = M2_activity(:,:,:,idx_lowV_LR);
for frame = 1 : size(M2_activity,3)
thisFrame = reshape(squeeze(M2_activity_lowV(:,:,frame,:)), [], sample_num*2);
ClassifierModel = fitclinear(thisFrame, keys, 'ObservationsIn', 'columns', 'Crossval', 'on');
Loss_experiment = kfoldLoss(ClassifierModel);
ShuffleModel = fitclinear(thisFrame, keys(randperm(length(keys))), 'ObservationsIn', 'columns', 'Crossval', 'on');
Loss_control = kfoldLoss(ShuffleModel);
Accuracy_data_lowV(frame, t) = (1 - Loss_experiment)*100;
Accuracy_shuffled_lowV(frame, t) = (1 - Loss_control)*100;
clear ClassifierModel ShuffleModel thisFrame
end
end
figure;
curve1 = stdshade(Accuracy_data_highV',0.4,'b');
ylim([30 100]);
hold on;
curve2 = stdshade(Accuracy_shuffled_highV',0.4,'r');
curve3 = stdshade(Accuracy_data_lowV',0.4,'c');
curve4 = stdshade(Accuracy_shuffled_lowV',0.4,'m');
legend([curve1, curve2, curve3, curve4], 'highV', 'highV shuffled', 'lowV', 'lowV shuffled');
xlabel('Frames in Response Window');
ylabel('Accuracy(%)');
TITLE = ['Linear Classifier for predicting the choice based on M2, ', mousename];
title(TITLE);
set(gca,'box','off');
set(gca,'tickdir','out');
hold off
clear curve1 curve2 curve3 curve4 TITLE
end
end