This repository has been archived by the owner on Feb 20, 2018. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 11
/
summarize_data.m
229 lines (211 loc) · 11.7 KB
/
summarize_data.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
% Copyright (C) 2010-2017, Raytheon BBN Technologies and contributors listed
% in the AUTHORS file in TASBE analytics package distribution's top directory.
%
% This file is part of the TASBE analytics package, and is distributed
% under the terms of the GNU General Public License, with a linking
% exception, as described in the file LICENSE in the TASBE analytics
% package distribution's top directory.
function experimentresults = summarize_data( colorModel, experiment, analysisParams, sampleresults)
%SUMMARIZE_DATA Given the data from each sample, summarize it into a single
% collection of statistics per condition
filenames = getInducerLevelsToFiles(experiment); % array of file names
n_conditions = numel(filenames);
n_channels = numel(getChannels(colorModel));
n_bins = get_n_bins(getBins(analysisParams));
n_components = getNumGaussianComponents(analysisParams);
% Combine all material together across replicates
bincounts = zeros(n_bins,n_conditions);
repcounts = zeros(n_bins,n_conditions);
tmp_expt_means = zeros(n_bins,n_conditions,n_channels);
tmp_expt_stds = zeros(n_bins,n_conditions,n_channels);
tmp_expt_popmeans = zeros(n_conditions,n_channels);
tmp_expt_popstds = zeros(n_conditions,n_channels);
tmp_expt_histograms = zeros(n_conditions,n_bins,n_channels);
tmp_expt_excluded = zeros(n_conditions, n_channels);
expt_nonexpressing = zeros(n_conditions);
tmp_expt_popcmeans = zeros(n_components,n_conditions,n_channels);
tmp_expt_popcstds = zeros(n_components,n_conditions,n_channels);
tmp_expt_popcweights = zeros(n_components,n_conditions,n_channels);
expt_plasmids = zeros(n_bins,n_conditions);
tmp_expt_stdofmeans = zeros(n_bins,n_conditions,n_channels);
tmp_expt_stdofstds = zeros(n_bins,n_conditions,n_channels);
tmp_expt_stdofpopmeans = zeros(n_conditions,n_channels);
tmp_expt_stdofpopstds = zeros(n_conditions,n_channels);
tmp_expt_stdofhistograms = zeros(n_conditions,n_bins,n_channels);
tmp_expt_stdofexcluded = zeros(n_conditions,n_channels);
expt_stdofnonexpressing = zeros(n_conditions);
tmp_expt_stdofpopcmeans = zeros(n_components,n_conditions,n_channels);
tmp_expt_stdofpopcstds = zeros(n_components,n_conditions,n_channels);
tmp_expt_stdofpopcweights = zeros(n_components,n_conditions,n_channels);
expt_stdofplasmids = zeros(n_bins,n_conditions);
max_events = 0; max_sample = [];
min_events = Inf; min_sample = [];
for i=1:n_conditions,
sampleset = [sampleresults{i}{:}];
% Only use bins with at least the minimum number of data elements
valid = [sampleset.BinCounts]>getMinValidCount(analysisParams);
% Add all valid bins to counts
repcounts(:,i) = sum(valid,2);
% pull sample data
sample_bc = [sampleset.BinCounts];
sample_means = reshape([sampleset.Means],n_bins,n_channels,numel(sampleset));
sample_stds = reshape([sampleset.StandardDevs],n_bins,n_channels,numel(sampleset));
sample_popmeans = reshape([sampleset.PopMeans],n_channels,numel(sampleset));
sample_popstds = reshape([sampleset.PopStandardDevs],n_channels,numel(sampleset));
sample_histograms = reshape([sampleset.Histograms],n_bins,n_channels,numel(sampleset));
sample_excluded = reshape([sampleset.Excluded],n_channels,numel(sampleset));
sample_nonexpressing = [sampleset.NonExpressing];
sample_popcmeans = reshape([sampleset.PopComponentMeans],n_components,n_channels,numel(sampleset));
sample_popcstds = reshape([sampleset.PopComponentStandardDevs],n_components,n_channels,numel(sampleset));
sample_popcweights = reshape([sampleset.PopComponentWeights],n_components,n_channels,numel(sampleset));
sample_plas = [sampleset.PlasmidEstimates];
sample_active = [sampleset.FractionActive];
% track min/max total number of events in a sample
for j=1:numel(sampleset)
event_count = sampleresults{i}{j}.Excluded(1) + sum(sampleresults{i}{j}.Histograms(:,1));
if event_count>max_events, max_events = event_count; max_sample = [i j]; end;
if event_count<min_events, min_events = event_count; min_sample = [i j]; end;
end
% first the bin statistics
for j=1:n_bins
bincounts(j,i) = sum(sample_bc(j,valid(j,:)));
% TODO: allow arithmetic statistics as well
% Note: Matlab handled all of these statistics using the general
% case of valud(j,:)>1; Octave has more safeties on statistics
% over higher dimensions, and in this case they backfire
if sum(valid(j,:))==0,
tmp_expt_means(j,i,:) = NaN;
tmp_expt_stdofmeans(j,i,:) = NaN;
tmp_expt_stds(j,i,:) = NaN;
tmp_expt_stdofstds(j,i,:) = NaN;
expt_plasmids(j,i) = NaN;
expt_stdofplasmids(j,i) = NaN;
expt_activity(j,i) = NaN;
expt_stdofactivity(j,i) = NaN;
else if sum(valid(j,:))==1,
tmp_expt_means(j,i,:) = sample_means(j,:,valid(j,:));
tmp_expt_stdofmeans(j,i,:) = 0;
tmp_expt_stds(j,i,:) = sample_stds(j,:,valid(j,:));
tmp_expt_stdofstds(j,i,:) = 0;
expt_plasmids(j,i) = sample_plas(j,valid(j,:));
expt_stdofplasmids(j,i) = 0;
expt_activity(j,i) = sample_active(j,valid(j,:));
expt_stdofactivity(j,i) = 0;
else
for k=1:n_channels,
tmp_expt_means(j,i,k) = geomean(sample_means(j,k,valid(j,:)));
tmp_expt_stdofmeans(j,i,k) = geostd(sample_means(j,k,valid(j,:)));
tmp_expt_stds(j,i,k) = geomean(sample_stds(j,k,valid(j,:)));
tmp_expt_stdofstds(j,i,k) = geostd(sample_stds(j,k,valid(j,:)));
end
expt_plasmids(j,i) = geomean(sample_plas(j,valid(j,:)));
expt_stdofplasmids(j,i) = geostd(sample_plas(j,valid(j,:)));
expt_activity(j,i) = geomean(sample_active(j,valid(j,:)));
expt_stdofactivity(j,i) = geostd(sample_active(j,valid(j,:)));
end
end
end
% and now the bulk statistics
for j=1:n_channels,
tmp_expt_popmeans(i,j) = geomean(sample_popmeans(j,:));
tmp_expt_stdofpopmeans(i,j) = geostd(sample_popmeans(j,:));
tmp_expt_popstds(i,j) = geomean(sample_popstds(j,:));
tmp_expt_stdofpopstds(i,j) = geostd(sample_popstds(j,:));
for k=1:n_bins,
tmp_expt_histograms(i,k,j) = geomean(sample_histograms(k,j,:));
tmp_expt_stdofhistograms(i,k,j) = geostd(sample_histograms(k,j,:));
end
tmp_expt_excluded(i,j) = geomean(sample_excluded(j,:));
tmp_expt_stdofexcluded(i,j) = geostd(sample_excluded(j,:));
end
expt_nonexpressing = geomean(sample_nonexpressing);
expt_stdofnonexpressing = geostd(sample_nonexpressing);
% and now the k-component gaussian population statistics
for j=1:n_components
if sum(valid(j,:))==0,
tmp_expt_popcmeans(j,i,:) = NaN;
tmp_expt_stdofpopcmeans(j,i,:) = NaN;
tmp_expt_popcstds(j,i,:) = NaN;
tmp_expt_stdofpopcstds(j,i,:) = NaN;
tmp_expt_popcweights(j,i,:) = NaN;
tmp_expt_stdofpopcweights(j,i,:) = NaN;
else if sum(valid(j,:))==1,
tmp_expt_popcmeans(j,i,:) = sample_popcmeans(j,:,valid(j,:));
tmp_expt_stdofpopcmeans(j,i,:) = 0;
tmp_expt_popcstds(j,i,:) = sample_popcstds(j,:,valid(j,:));
tmp_expt_stdofpopcstds(j,i,:) = 0;
tmp_expt_popcweights(j,i,:) = sample_popcweights(j,:,valid(j,:));
tmp_expt_stdofpopcweights(j,i,:) = 0;
else
for k=1:n_channels,
tmp_expt_popcmeans(j,i,k) = geomean(sample_popcmeans(j,k,valid(j,:)));
tmp_expt_stdofpopcmeans(j,i,k) = geostd(sample_popcmeans(j,k,valid(j,:)));
tmp_expt_popcstds(j,i,k) = geomean(sample_popcstds(j,k,valid(j,:)));
tmp_expt_stdofpopcstds(j,i,k) = geostd(sample_popcstds(j,k,valid(j,:)));
tmp_expt_popcweights(j,i,k) = mean(sample_popcweights(j,k,valid(j,:)));
tmp_expt_stdofpopcweights(j,i,k) = std(sample_popcweights(j,k,valid(j,:)));
end
end
end
end
end
% Test for sanity on variation in sample/sample event counts - should be less than 10x
if max_events/min_events > 10
warning('TASBE:SampleSize','High variation in events per sample:\n Condition %i, sample %i = %i\n Condition %i, sample %i = %i',...
max_sample(1), max_sample(2), max_events, min_sample(1), min_sample(2), min_events);
end
% Fill in per-channel outputs
expt_means = cell(n_channels,2); expt_stds = cell(n_channels,2);
expt_stdofmeans = cell(n_channels,2); expt_stdofstds = cell(n_channels,2);
expt_popmeans = cell(n_channels,2); expt_popstds = cell(n_channels,2);
expt_histograms = cell(n_channels,2); expt_excluded = cell(n_channels,2);
expt_stdofpopmeans = cell(n_channels,2); expt_stdofpopstds = cell(n_channels,2);
expt_stdofhistograms = cell(n_channels,2); expt_stdofexcluded = cell(n_channels,2);
expt_popcmeans = cell(n_channels,2); expt_popcstds = cell(n_channels,2); expt_popcweights = cell(n_channels,2);
expt_stdofpopcmeans = cell(n_channels,2); expt_stdofpopcstds = cell(n_channels,2); expt_stdofpopcweights = cell(n_channels,2);
% First entry of each output table is the Channel, second is the statistic array
for i=1:n_channels,
expt_means{i,1} = getChannel(colorModel,i);
expt_stds{i,1} = getChannel(colorModel,i);
expt_stdofmeans{i,1} = getChannel(colorModel,i);
expt_stdofstds{i,1} = getChannel(colorModel,i);
expt_means{i,2} = tmp_expt_means(:,:,i);
expt_stds{i,2} = tmp_expt_stds(:,:,i);
expt_stdofmeans{i,2} = tmp_expt_stdofmeans(:,:,i);
expt_stdofstds{i,2} = tmp_expt_stdofstds(:,:,i);
expt_popmeans{i,1} = getChannel(colorModel,i);
expt_popstds{i,1} = getChannel(colorModel,i);
expt_histograms{i,1} = getChannel(colorModel,i);
expt_excluded{i,1} = getChannel(colorModel,i);
expt_stdofpopmeans{i,1} = getChannel(colorModel,i);
expt_stdofpopstds{i,1} = getChannel(colorModel,i);
expt_stdofhistograms{i,1} = getChannel(colorModel,i);
expt_stdofexcluded{i,1} = getChannel(colorModel,i);
expt_popmeans{i,2} = tmp_expt_popmeans(:,i);
expt_popstds{i,2} = tmp_expt_popstds(:,i);
expt_histograms{i,2} = tmp_expt_histograms(:,:,i);
expt_excluded{i,2} = tmp_expt_excluded(:,i);
expt_stdofpopmeans{i,2} = tmp_expt_stdofpopmeans(:,i);
expt_stdofpopstds{i,2} = tmp_expt_stdofpopstds(:,i);
expt_stdofhistograms{i,2} = tmp_expt_stdofhistograms(:,:,i);
expt_stdofexcluded{i,2} = tmp_expt_stdofexcluded(:,i);
expt_popcmeans{i,1} = getChannel(colorModel,i);
expt_popcstds{i,1} = getChannel(colorModel,i);
expt_popcweights{i,1} = getChannel(colorModel,i);
expt_stdofpopcmeans{i,1} = getChannel(colorModel,i);
expt_stdofpopcstds{i,1} = getChannel(colorModel,i);
expt_stdofpopcweights{i,1} = getChannel(colorModel,i);
expt_popcmeans{i,2} = tmp_expt_popcmeans(:,:,i);
expt_popcstds{i,2} = tmp_expt_popcstds(:,:,i);
expt_popcweights{i,2} = tmp_expt_popcweights(:,:,i);
expt_stdofpopcmeans{i,2} = tmp_expt_stdofpopcmeans(:,:,i);
expt_stdofpopcstds{i,2} = tmp_expt_stdofpopcstds(:,:,i);
expt_stdofpopcweights{i,2} = tmp_expt_stdofpopcweights(:,:,i);
end
experimentresults = ExperimentResults(experiment, analysisParams, repcounts, bincounts, ...
expt_means, expt_stds, expt_plasmids, expt_activity, ...
expt_popmeans, expt_popstds, expt_histograms, expt_excluded, expt_nonexpressing, ...
expt_popcmeans, expt_popcstds, expt_popcweights, ...
expt_stdofmeans, expt_stdofstds, expt_stdofplasmids, expt_stdofactivity, ...
expt_stdofpopmeans, expt_stdofpopstds, expt_stdofhistograms, expt_stdofexcluded, expt_stdofnonexpressing, ...
expt_stdofpopcmeans, expt_stdofpopcstds, expt_stdofpopcweights);