-
Notifications
You must be signed in to change notification settings - Fork 1
/
analyze_etkinlab_siteeffects.m
258 lines (209 loc) · 8.87 KB
/
analyze_etkinlab_siteeffects.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
function results = analyze_etkinlab_siteeffects()
% Compute and return conditional correlations, nuisance correlations, and NSR on all subjects x sites.
datadir = 'data';
studydata = load(fullfile(datadir,'HCData4SiteEfectTests.mat'));
studydata.noise = load(fullfile(datadir,'noisePCs.mat'));
studydata.yeo_order = load(fullfile(datadir,'community_ordering'));
reorder_community = studydata.yeo_order.orderbycommunities;
studydata.compcor1 = studydata.noise.noisePCs.noisePCs(1:20,:,:);
studydata.compcor2 = studydata.noise.noisePCs.noisePCs(21:end,:,:);
results = struct();
results.Site1 = {};
results.Site2 = {};
observed_correlation = [];
denoised_correlation = [];
denoised_correlation2 = [];
nuisance_correlation = [];
nsr = [];
sitelabels = zeros(44,1); sitelabels(1:20) = 1; sitelabels(21:44) = 2;
includeROIs1 = find(sum(squeeze(abs(std(studydata.signalsSite1,[],3))>1e-2),1)==length(studydata.subIDsSite1));
includeROIs2 = find(sum(squeeze(abs(std(studydata.signalsSite2,[],3))>1e-2),1)==length(studydata.subIDsSite2));
includeROIs = intersect(includeROIs1,includeROIs2);
[loca locb] = ismember(reorder_community,includeROIs);
choose_community = reorder_community<=3;
compcor_dims = size(studydata.compcor1);
%Y = get_shared_nuisance(squeeze(mean(studydata.signalsSite1,2))');
Y = [];
for ii=1:length(studydata.subIDsSite1)
ii
X = squeeze(studydata.signalsSite1(ii,reorder_community(loca(:)),:))';
Y = cat(2,Y,squeeze(studydata.compcor1(ii,:,:)));
try
observed_correlation(ii,:,:) = standard_correlation(X);
results.Site1{ii} = conditional_correlation(X,Y);
denoised_correlation(ii,:,:) = results.Site1{ii}.corr;
denoised_correlation2(ii,:,:) = standard_correlation_sn(X);
nuisance_correlation(ii,:,:) = results.Site1{ii}.nuisance; nsr(ii) = results.Site1{ii}.NSR;
catch me
disp(me)
results.Site1{ii} = {};
end
% if(ii==1)
% demo_conditional_correlation(X,Y);
% end
end
baseidx = size(denoised_correlation,1);
%Y = get_shared_nuisance(squeeze(mean(studydata.signalsSite2,2))');
Y = [];
for ii=1:length(studydata.subIDsSite2)
X = squeeze(studydata.signalsSite2(ii,reorder_community(loca(:)),:))';
Y = cat(2,Y,squeeze(studydata.compcor2(ii,:,:)));
try
observed_correlation(baseidx+ii,:,:) = standard_correlation(X);
results.Site2{ii} = conditional_correlation(X,Y);
denoised_correlation(baseidx+ii,:,:) = results.Site2{ii}.corr;
denoised_correlation2(baseidx+ii,:,:) = standard_correlation_sn(X);
nuisance_correlation(baseidx+ii,:,:) = results.Site2{ii}.nuisance;
nsr(baseidx+ii) = results.Site2{ii}.NSR;
catch me
disp(me)
results.Site2{ii} = {};
end
% if(ii==1)
% demo_conditional_correlation(X,Y);
% end
end
results.denoised = denoised_correlation;
results.denoised2 = denoised_correlation2;
results.nuisance = nuisance_correlation;
results.observed = observed_correlation;
results.nsr = nsr;
if(exist('brewermap'))
colormapfun = @()(flipud(brewermap(length(colormap),'RdYlBu')));
close all;
else
colormapfun = @winter;
end
addpath(genpath('~/MATLAB/packages/spreadFigures/tightfit/'));
figure(1);
%set(gcf,'Position',[10 450 1200 300]);
set(gcf,'Position',[10 450 850 850]);
subplot(2,2,1);
site_effect{1} = detect_site_effect(results.observed,sitelabels);
imagesc(real(site_effect{1}.similarity));
colormap(colormapfun()); colorbar; axis image;
title('Similarity Matrix (Observed)','fontsize',24);
xlabel(sprintf('(wit,bet,rat) = (%.2f, %.2f, %.2f)', ...
site_effect{1}.within, ...
site_effect{1}.between, ...
site_effect{1}.ratio));
set(gca,'fontsize',16);
subplot(2,2,2);
site_effect{2} = detect_site_effect(results.nuisance,sitelabels);
imagesc(real(site_effect{2}.similarity));
colormap(colormapfun()); colorbar; axis image;
title('Similarity Matrix (Nuisance)','fontsize',24);
xlabel(sprintf('(wit,bet,rat) = (%.2f, %.2f, %.2f)', ...
site_effect{2}.within, ...
site_effect{2}.between, ...
site_effect{2}.ratio));
set(gca,'fontsize',16);
subplot(2,2,3);
site_effect{3} = detect_site_effect(results.denoised,sitelabels);
imagesc(site_effect{3}.similarity);
colormap(colormapfun()); colorbar; axis image;
title('Similarity Matrix (Denoised)','fontsize',24);
xlabel(sprintf('(wit,bet,rat) = (%.2f, %.2f, %.2f)', ...
site_effect{3}.within, ...
site_effect{3}.between, ...
site_effect{3}.ratio));
set(gca,'fontsize',16);
subplot(2,2,4);
site_effect{4} = detect_site_effect(results.denoised2,sitelabels);
imagesc(site_effect{4}.similarity);
colormap(colormapfun()); colorbar; axis image;
title('Similarity Matrix (SN)','fontsize',24);
xlabel(sprintf('(wit,bet,rat) = (%.2f, %.2f, %.2f)', ...
site_effect{4}.within, ...
site_effect{4}.between, ...
site_effect{4}.ratio));
set(gca,'fontsize',16);
% if(exist('tightfig'))
% tightfig;
% end
results.site_effect = site_effect;
exportfun = @(fname)(print('-dpng','-r600',fname));
fname = ['tmp' filesep datestr(now,'dd-mmm-yyyy-HHMM')];
if(~exist(fname,'dir'))
mkdir(fname)
end
exportfun(fullfile(fname,[mfilename '1']));
% figure;
% % stem([1:length(nsr)],nsr); hold on;
% % scatter([1:length(nsr)],nsr,25); hold off;
% % ylim([0 1.1]); xlim([0 length(nsr)+2]);
% % set(gca,'fontsize',24);
% % colormap(colormapfun()); axis image; colorbar;
% boxplot(nsr,'group',sitelabels,'color',sitelabels)
% title('Nuisance to Signal Ratio by Site')
% exportfun(fullfile(fname,[mfilename '2']));
%
% results(1).site_effect = site_effect;
% CAUTION
% These diagnostics will not work until signal has been cleaned.
%
% X1 = permute(studydata.signalsSite1,[3 2 1]);
% X2 = permute(studydata.signalsSite2,[3 2 1]);
% results.diagnostic1 = get_influence_diagnostic(X1);
% results.diagnostic2 = get_influence_diagnostic(X2);
%
% results.diagnostic3 = get_influence_diagnostic(cat(3,X1,X2));
%
%
% save('~/COMET/adim-preprocessed/others/HCData4SiteEffect_Decomposition_CompCorNuisance','results');
end
function output = conditional_correlation(X,Y)
% Only uses usual column standardize (i.e. correlation)
output = struct();
[Sigma results] = covariance.conditional_sample_covariance_separate(X, ...
struct('verbose',false,...
'nuisance',Y) ...
);
output.corr = results.Sigma;
output.nuisance = results.nCorr;
output.corr2 = covariance.mle_sample_covariance(results.X_perpY, ...
struct('standardize', false));
output.NSR = results.NSR;
end
function Sigma = standard_correlation(X)
% Usual standard correlation matrix
[Sigma results] = covariance.mle_sample_covariance(X, ...
struct('standardize',false));
end
function Sigma = standard_correlation_sn(X)
% Automatically applies row-first successive norm
%standardize.successive_normalize(X');
[Xnew] = standardize.successive_normalize(X');
[Sigma results] = corr(Xnew');
% [Sigma results] = covariance.mle_sample_covariance(X,
% struct('standardize',true));
%
end
function site_effect = detect_site_effect(X,sitelabels)
corrfun = @(X)(corr(X)); %@(X)(covariance.rank_sample_covariance(X(:,1:min(50,size(X,2)))));
upper_idx = find(reshape(triu(ones(size(X,2), size(X,3)),1), [1 size(X,2)^2]));
X = reshape(X,[size(X,1) size(X,2)*size(X,3)]);
tmpSimilarity = corrfun(X(:,upper_idx)');
[hom sep mw] = compareWithinAndBetweenGroupsSim(tmpSimilarity,sitelabels);
site_effect.similarity = tmpSimilarity;
site_effect.within = hom;
site_effect.between = sep;
site_effect.stat = mw;
site_effect.ratio = (hom-sep)/(hom+sep);
end
function [nuisance] = get_shared_nuisance(Y);
% Y is time-series x subjects
Yz = zscore(Y')';
[U S V] = svd(Y);
nfactors = min(size(Y,2),5);
nuisance = U(1:nfactors,:);
nuisance = reshape(nuisance,[size(Y,1) nfactors]);
end
function output = get_influence_diagnostic(X)
output = struct();
[Sigmasb Sigmas] = covjackknife(X,[1 2 3]);
output.nansubjects = isnan(squeeze(sum(sum(Sigmas,1),2)));
Sigmasb(isnan(Sigmasb)) = 0;
Sigmas(isnan(Sigmas)) = 0;
[output.influence output.norminfluence] = influence(Sigmasb,Sigmas);
end