-
Notifications
You must be signed in to change notification settings - Fork 0
/
cosmo.m~
315 lines (283 loc) · 13.6 KB
/
cosmo.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
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
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
function varargout = cosmo(what,varargin)
%hello world
% Define common stuff here
baseDir = fileparts(which('cosmo.m'));
dataDir = fullfile(baseDir,'data');
stimLow = 1; % Define stimulus range
stimHigh = 5;
dStim = 0.25; % define stimulus steps
stims = stimLow:dStim:stimHigh;
numStim = numel(stims);
numNeuron = 1000; % number of neurons altogether
numRun = 8; % 8 runs
numRep = 10; % 10 repetitions per run, (80) overall
switch what
case 'GEN_tunedPopulation'
% generate tuning population of neurons
% usage: cosmo('GEN_population','numNeuron',1000,'numStim',3);
numPrefs = 5;
plotFig = 1;
vararginoptions(varargin,{'numNeuron','numPrefs','plotFig','scale','offset','sigma'});
% determine preferred tuning and variance per neuron
prefDir = randi(5,[numNeuron,1]);
sigma = 7*rand(numNeuron,1)+0.5; % 0.4 former
scale = rand([numNeuron,1]); % 1 former
offset = rand([numNeuron,1])*0.75 + 0.5; % 0.5 former
% organised preferred direction - pref: 1,2,...,numPrefs
TC = @(scale,stim,prefDir,sigma,offset)...
scale .* exp(-((stim-prefDir).^2)./sigma) + offset;
for i = 1:numNeuron
tuning(i,:) = TC(scale(i),stims,prefDir(i),sigma(i),offset(i));
end
rescale = max(max(tuning));
tuning = tuning./rescale;
scale = scale./rescale;
offset = offset./rescale;
% gIndep = tuning.*1.5;
if plotFig==1 % optional plotting of tuning functions across neurons
figure;
hold on;
for i=1:numPrefs
inds = prefDir==i;
idx = find(inds==1);
subplot(1,numPrefs,i)
%plot(stims,TC(scale(inds)),stims,prefDir(inds),sigma(inds),offset(inds)));
end
end
% save the tuning matrix (numNeuron x numStim)
save(fullfile(dataDir,sprintf('tunMatrix_%dneurons_%dstim',numNeuron,numStim)),...
'tuning','prefDir','sigma','scale','offset');
case 'GEN_LIF'
% define default parameters for LIFModel
gShared = 0.003; % shared noise
gIndep = 0.008; % independent noise
gAnat = 0.001;
plotOn = 0;
stimDur = 2; % in seconds
dT = 0.001; % time increment
numRun = 8; % 8 runs
numRep = 10; % 10 repetitions per run, (80) overall
spikeScale = 80; % in Hz
TC = @(scale,stim,prefDir,sigma,offset)...
scale * exp(-((stim-prefDir).^2)./sigma) + offset;
popType = 'mixture';
vararginoptions(varargin,{'stimRate','gShared','gIndep','plotOn','numNeuron','numStim','dt','stimDur','numRep','spikeScale','popType'});
TT=[]; % initialise for storage (spikes across neurons / stimuli)
% load the correct tuning matrix
D = load(fullfile(dataDir,sprintf('tunMatrix_%dneurons_%dstim',numNeuron,numStim)));
switch popType
case 'mixture'
anatVec = repmat([1; -1],numNeuron,1);
case 'positive'
anatVec = repmat([1; 1],numNeuron,1);
end
for t=1:numStim
for r=1:numRun
for rep=1:numRep
gSharedVec = sharedNoise(gShared,dT,stimDur); % same across neurons
for n=1:numNeuron
% 1) determine spike rate based on tuning (pull
% from params
resp = TC(D.scale(n),stims(t),D.prefDir(n),D.sigma(n),D.offset(n));
% resp = D.tuning(n,t);
spkRate = resp*spikeScale;
% 2) generate spikes
[spkInds,spkVec] = genSpikes(stimDur,spkRate,dT);
% 3) add anatOff
anatSign = anatVec(n);
% 4) run the LIFModel
%T.spikes{1} = LIFModel(spkInds,spkVec,gSharedVec,0,gAnat,anatSign,dT,stimDur,plotOn);
T.spikes{1} = LIFModel(spkInds,spkVec,gSharedVec,gIndep,gAnat,anatSign,dT,stimDur,plotOn);
T.spikeNum = numel(T.spikes{1});
T.neuron = n;
T.numRun = r;
T.numRep = rep;
T.prefDir = D.prefDir(n);
T.stimDir = stims(t);
T.anatSign = anatSign;
TT = addstruct(TT,T);
clear spkVec spkRate spkInds;
end
end
fprintf('Generated all neurons for stimulus: %d/%d runs %d/%d\n',t,numStim,r,numRun);
end
end
figure;
lineplot(TT.stimDir,TT.spikeNum,'split',TT.prefDir,'style_thickline');
xlabel('Direction'); ylabel('Spike number'); title('Responses split by preferred direction');
% save the outputs
save(fullfile(dataDir,sprintf('LIF_%dneurons_%dstim_%sPopulation',numNeuron,numStim,popType)),'-struct','TT');
case 'PLOT_population'
% plot the population
popType = 'mixture';
vararginoptions(varargin,{'numNeuron','numStim','popType'});
T = load(fullfile(dataDir,sprintf('LIF_%dneurons_%dstim_%sPopulation',numNeuron,numStim,popType)));
for l=1:numStim
legLab{l} = sprintf('stim-%d',l);
end
figure
plt.hist(T.spikeNum,'split',T.prefDir);
ylabel('number of spikes');
% extract variance and mean
T1=tapply(T,{'neuron','prefDir','stimDir'},{'spikeNum','mean','name','spikeNum_mean'},...
{'spikeNum','var','name','spikeNum_var'});
% plot responses in dependence of preferred - presented stimulus
figure(1)
subplot(1,2,1)
%lineplot(abs(T1.prefDir-T1.stimDir),T1.spikeNum_mean,'split',T1.prefDir,...
% 'leg',legLab,'style_thickline','markertype','o','markersize',12);
plt.line(abs(T1.prefDir-T1.stimDir),T1.spikeNum_mean,'split',T1.prefDir,...
'leg',legLab,'markertype','o','markersize',12);
xlabel('Preferred-presented stimulus');
ylabel('Mean spike number response');
subplot(1,2,2)
% calculate the variance
% lineplot(abs(T1.prefDir-T1.stimDir),T1.spikeNum_var,'split',T1.prefDir,...
% 'leg',legLab,'style_thickline','markertype','o','markersize',12);
plt.line(abs(T1.prefDir-T1.stimDir),T1.spikeNum_var,'split',T1.prefDir,...
'leg',legLab,'markertype','o','markersize',12);
ylabel('Variance in spike number across trials');
% plot stuff
figure(2)
barplot(abs(T1.prefDir-T1.stimDir),T1.spikeNum_var,'split',T1.prefDir);
case 'CALC_corr_dprime'
popType = 'mixture';
vararginoptions(varargin,{'numNeuron','numStim','popType'});
T = load(fullfile(dataDir,sprintf('LIF_%dneurons_%dstim_%sPopulation',numNeuron,numStim,popType)));
% extract variance and mean
T1=tapply(T,{'neuron','prefDir','stimDir'},{'spikeNum','mean','name','spikeNum_mean'},...
{'spikeNum','var','name','spikeNum_var'});
DD=[];
% subtract the condition mean across trials
NN=[];
for sd=1:numStim
for n=1:numNeuron
N1=getrow(T,T.neuron==n&T.stimDir==sd);
N2=getrow(T1,T1.neuron==n&T1.stimDir==sd);
N1.spikeNum=N1.spikeNum-N2.spikeNum_mean;
NN=addstruct(NN,N1);
end
end
T2=tapply(NN,{'neuron','prefDir','stimDir'},{'spikeNum','mean','name','spikeNum_mean'},...
{'spikeNum','var','name','spikeNum_var'});
for i=1:numNeuron
for j=i:numNeuron
D.corrMean = corr(T1.spikeNum_mean(T1.neuron==i),T1.spikeNum_mean(T1.neuron==j));
D.corrVar = corr(T2.spikeNum_var(T2.neuron==i),T2.spikeNum_var(T2.neuron==j));
for c=1:numStim
t1=getrow(T,T.neuron==i & T.stimDir==c);
t2=getrow(T,T.neuron==j & T.stimDir==c);
D.dprime(1,c)=dprime(t1.spikeNum,t2.spikeNum);
end
D.neuron1 = i;
D.neuron2 = j;
D.sameNeuron = double(i==j);
pref1 = T1.prefDir(T1.neuron==i);
pref2 = T1.prefDir(T1.neuron==j);
anat1 = T.anatSign(T.neuron==i);
anat2 = T.anatSign(T.neuron==i);
D.prefDir1 = pref1(1);
D.prefDir2 = pref2(1);
D.prefSame = double(D.prefDir1==D.prefDir2);
D.anatSign1 = anat1(1);
D.anatSign2 = anat2(1);
DD = addstruct(DD,D);
Rm(i,j)=D.corrMean;
Rv(i,j)=D.corrVar;
end
fprintf('Calc corr pairs:\tneuron %d/%d\n',i,numNeuron);
end
save(fullfile(dataDir,sprintf('corr_neuronPairs_%dneurons_%sPopulation',numNeuron,popType)),'-struct','DD');
case 'CALC_classify'
popType = 'mixture';
vararginoptions(varargin,{'numNeuron','numStim','popType'});
T = load(fullfile(dataDir,sprintf('LIF_%dneurons_%dstim_%sPopulation',numNeuron,numStim,popType)));
% add partVec
T1=tapply(T,{'stimDir','numRun','neuron'},{'spikeNum','mean'});
% rearrange
for i=1:length(unique(T1.numRun))
tmp = getrow(T1,T1.numRun==i);
[indx,j,k] = pivottable([tmp.stimDir],[tmp.neuron],[tmp.spikeNum],'mean');
data(:,:,i) = indx;
end
% choose subsets of neurons
nNeuron=48;
switch popType
case 'mixture'
% positive / negative / mixed corr
indxN(:,1)=randsample(unique(T.neuron(T.anatSign==1)),nNeuron);
indxN(:,2)=randsample(unique(T.neuron(T.anatSign==-1)),nNeuron);
indxN(:,3)=[randsample(indxN(:,1),nNeuron/2); randsample(indxN(:,2),nNeuron/2)]';
case 'positive'
indxN(:,1)=randsample(unique(T.neuron),nNeuron)';
end
for s=1:size(indxN,2)
for i=1:2 % give all stimuli or only 5
if i==1 % then give all stimuli
data_sub = data(:,indxN(:,s),:);
T_sub = getrow(T,ismember(T.neuron,indxN(:,s)));
acc(s,i) = nn_classifier(data_sub,T_sub.spikeNum,T_sub.numRun,T_sub.stimDir,T_sub.numRep);
else
indxS = ismember(stims,[1:numStim]);
data_sub = data(indxS,indxN(:,s),:);
T_sub = getrow(T,ismember(T.stimDir,T.prefDir) & ismember(T.neuron,indxN(:,s)));
acc(s,i) = nn_classifier(data_sub,T_sub.spikeNum,T_sub.numRun,T_sub.stimDir,T_sub.numRep);
end
% submit to classifier, distance calculation
end
end
% dist = rsa_distanceLDC(T1.spikeNum,T1.numRun,T1.stimDir);
% % plot distances
% figure
% imagesc(rsa_squareRDM(dist));
keyboard;
% save classifier results
case 'PLOT_corr'
popType='mixture';
vararginoptions(varargin,{'numNeuron','popType'});
T = load(fullfile(dataDir,sprintf('corr_neuronPairs_%dneurons_%sPopulation',numNeuron,popType)));
figure
scatterplot(T.corrMean,T.corrVar,'split',T.prefSame,'leg',{'tuning different','tuning same'},'subset',T.sameNeuron==0);
%plt.scatter(T.corrMean,T.corrVar,'split',T.prefSame,'leg',{'tuning same','tuning different'});
xlabel('Signal correlation');
ylabel('Noise correlation');
title('Correlation structure across neuron pairs');
% make into matrix
M_mean = rsa_squareIPM(T.corrMean');
M_var = rsa_squareIPM(T.corrVar');
figure
subplot(1,3,1)
imagesc(M_mean);
title(sprintf('Signal correlation across neuron pairs - %s population',popType));
subplot(1,3,2)
imagesc(M_var);
title('Noise correlation across neuron pairs');
subplot(1,3,3)
imagesc(M_mean-M_var);
title('Difference signal-noise correlation');
figure
subplot(2,2,1)
plt.scatter(T.corrMean,T.corrVar,'subset',T.sameNeuron==0 & T.prefSame==0);
xlabel('Signal correlation'); ylabel('Noise correlation');
title(sprintf('Neurons with diff preferred direction - %s population',popType));
subplot(2,2,2)
plt.scatter(T.corrMean,T.corrVar,'subset',T.sameNeuron==0 & T.prefSame==1);
xlabel('Signal correlation'); ylabel('Noise correlation');
title('Neurons with same preferred direction');
subplot(2,2,3)
plt.hist(T.corrMean,'subset',T.sameNeuron==0);
hold on;
drawline(mean(T.corrMean(T.sameNeuron==0)),'dir','vert','color',[1 0 0]);
title('Distribution of signal correlation');
subplot(2,2,4)
plt.hist(T.corrVar,'subset',T.sameNeuron==0);
hold on;
drawline(mean(T.corrVar(T.sameNeuron==0)),'dir','vert','color',[1 0 0]);
title('Distribution of noise correlation');
case 'CALC_fisherInfo'
numRpts = numrun*numRep;
% [FI_corr,pop_dprime] = fisherInfo(dataDir,numNeuron,numRpts,numStim);
otherwise
fprintf('No such case\n');
end
% Local functions