Merge pull request #1493 from softmatterlab/alpha2_MultiplexPipelines

Alpha2 - multiplex pipelines

Author: giovannivolpe

braph2genesis/measures/_Distance.gen.m

@@ -39,13 +39,16 @@
 
 distance = cell(g.layernumber(), 1);
 connectivity_type =  g.getConnectivityType(g.layernumber());
-parfor li = 1:1:g.layernumber()
-
-    Aii = A{li, li};
-    connectivity_layer = connectivity_type(li, li);
+connectivity_type = diag(connectivity_type);
+for li = 1:g.layernumber()
+    Aii_tmp{li} = A{li, li}; %#ok<AGROW>
+end
+parfor li = 1:g.layernumber()
+    Aii = Aii_tmp{li};
+    connectivity_layer = connectivity_type(li);
 
     if connectivity_layer == Graph.WEIGHTED  % weighted graphs
-        distance(li) = {m.getWeightedCalculation(Aii)};
+        distance(li) = {m.getWeightedCalculation(Aii)}; %#ok<PFBNS> 
     else  % binary graphs
         distance(li) = {m.getBinaryCalculation(Aii)};
     end
@@ -54,7 +57,7 @@
 value = distance;
 
 %% ¡methods!
-function weighted_distance = getWeightedCalculation(m, A)
+function weighted_distance = getWeightedCalculation(~, A)
     %GETWEIGHTEDCALCULATION calculates the distance value of a weighted adjacency matrix.
     %
     % WEIGHTED_DISTANCE = GETWEIGHTEDCALCULATION(M, A) returns the value of the
@@ -77,9 +80,9 @@
 
             for v = V
                 T = find(L1(v, :)); % neighbours of shortest nodes
-                [d, wi] = min([D(u, T);D(u, v)+L1(v, T)]);
+                [d, ~] = min([D(u, T);D(u, v)+L1(v, T)]);
                 D(u, T) = d; % smallest of old/new path lengths
-                ind = T(wi==2); % indices of lengthened paths
+                % ind = T(wi==2); % indices of lengthened paths
             end
 
             minD = min(D(u, S));
@@ -92,7 +95,7 @@
     end
     weighted_distance = D;
 end
-function binary_distance = getBinaryCalculation(m, A)
+function binary_distance = getBinaryCalculation(~, A)
     %GETBINARYCALCULATION calculates the distance value of a binary adjacency matrix.
     %
     % BINARY_DISTANCE = GETBINARYCALCULATION(A) returns the value of the

braph2genesis/measures/_PathLength.gen.m

@@ -28,9 +28,9 @@
 %%% ¡prop! 
 rule (parameter, OPTION) is the path length algorithm
 %%%% ¡settings!
-{'subgraphs' 'harmonic' 'pl_default'}
+{'subgraphs' 'harmonic' 'mean'}
 %%%% ¡default!
-'pl_default'
+'harmonic'
 
 %% ¡props_update!
 
@@ -56,23 +56,22 @@
             for u = 1:1:N
                 Du = distance_layer(:, u);
                 path_length_layer(u) = mean(Du(Du~=Inf & Du~=0));
-            end 
-        case {'harmonic'}
+            end
+        case {'mean'}
             for u = 1:1:N
                 Du = distance_layer(:, u);
-                path_length_layer(u) = harmmean(Du(Du~=0));
+                path_length_layer(u) = mean(Du(Du~=0));
             end
-        otherwise  % 'default'
+        otherwise  % 'harmonic' 'default'
             for u = 1:1:N
                 Du = distance_layer(:, u);
-                path_length_layer(u) = mean(Du(Du~=0));
+                path_length_layer(u) = harmmean(Du(Du~=0));
             end
     end 
     path_length(li) = {path_length_layer}; % node Inf corresponds to isolated nodes
 end
 value = path_length;
 
-
 %% ¡tests!
 
 %%% ¡test!
@@ -86,7 +85,7 @@
     0   0   .1  0
     ];
 
-known_path_length = {[2 4/3 4/3 2]'};
+known_path_length = {[18/11 18/15 18/15 18/11]'};
 
 g = GraphBU('B', B);
 path_length = PathLength('G', g).get('M');
@@ -109,8 +108,8 @@
 thresholds = [0 1];
 
 known_path_length = { ...
-    [2   4/3  4/3  2]'
-    [Inf Inf  Inf  Inf]'
+    [18/11 18/15 18/15 18/11]'
+    [Inf   Inf   Inf   Inf]'
     };
 
 g = MultigraphBUT('B', B, 'THRESHOLDS', thresholds);
@@ -141,8 +140,8 @@
 B = { A11  A22};
 
 known_path_length = {
-                    [2 4/3 4/3 2]'
-                    [2 4/3 4/3 2]'
+                    [18/11 18/15 18/15 18/11]'
+                    [18/11 18/15 18/15 18/11]'
                     };
 
 g = MultiplexBU('B', B);

braph2genesis/measures/_PathLengthAv.gen.m

@@ -27,14 +27,23 @@
 %% ¡props_update!
 
 %%% ¡prop!
-M (result, cell) is the path length.
+M (result, cell) is the average path length.
 %%%% ¡calculate!
 g = m.get('G');  % graph from measure class
 
-path_length = calculateValue@PathLength(m, prop);            
+path_length = calculateValue@PathLength(m, prop);
 path_length_av = cell(g.layernumber(), 1);
+path_length_rule = m.get('rule');
 parfor li = 1:1:length(path_length_av)
-    path_length_av(li) = {mean(path_length{li})};
+    switch lower(path_length_rule)
+        case {'subgraphs'}
+            player = path_length{li};
+            path_length_av(li) = {mean(player(player~=Inf))};
+        case {'mean'}
+            path_length_av(li) = {mean(path_length{li})};
+        otherwise  % 'harmonic' 'default'
+            path_length_av(li) = {harmmean(path_length{li})};
+    end
 end
 value = path_length_av;
 
@@ -51,7 +60,7 @@
     0   0   .1  0
     ];
 
-known_path_length_av = {mean([2 4/3 4/3 2])};
+known_path_length_av = {mean([18/11 18/15 18/15 18/11])};
 
 g = GraphBU('B', A);
 path_length_av = PathLengthAv('G', g).get('M');
@@ -79,8 +88,8 @@
 A = {A11  A22};
 
 known_path_length_av = {
-                       mean([2 4/3 4/3 2])
-                       mean([2 4/3 4/3 2])
+                       mean([18/11 18/15 18/15 18/11])
+                       mean([18/11 18/15 18/15 18/11])
                        };
 
 g = MultiplexBU('B', A);

braph2genesis/measures/_SmallWorldness.gen.m

@@ -37,9 +37,9 @@
 end
 L = g.layernumber();
 
+path_length_rule = m.get('rule');
 clustering_av = ClusteringAv('G', g).get('M');
 path_length_av = calculateValue@PathLengthAv(m, prop);
-path_length_rule = m.get('rule');
 
 M = 100;  % number of random graphs
 clustering_av_random = cell(1, M);

braph2genesis/pipelines/connectivity-functional multiplex group_average/_AnalyzeGroup_CON_FUN_MP_GA_BUD.gen.m

@@ -0,0 +1,146 @@
+%% ¡header!
+AnalyzeGroup_CON_FUN_MP_GA_BUD < AnalyzeGroup (a, graph analysis with connectivity and functional multiplex data of fixed density) is a graph analysis using connectivity and functional multiplex data of fixed density.
+
+%% ¡description!
+This graph analysis uses connectivity and functional multiplex data of 
+fixed density and analyzes them using binary undirected graphs.
+
+%%% ¡seealso!
+AnalyzeGroup_CON_FUN_MP_WU, SubjectCON_FUN_MP, MultiplexBUD.
+
+%% ¡props!
+
+%%% ¡prop!
+REPETITION(parameter, scalar) is the number of repetitions for functional data
+%%%% ¡default!
+1
+
+%%% ¡prop!
+FREQUENCYRULEMIN(parameter, scalar)is the minimum frequency value for functional data
+%%%% ¡default!
+0
+
+%%% ¡prop!
+FREQUENCYRULEMAX(parameter, scalar)is the maximum frequency value for functional data
+%%%% ¡default!
+Inf
+
+%%% ¡prop!
+CORRELATION_RULE (parameter, option) is the correlation type for functional data.
+%%%% ¡settings!
+Correlation.CORRELATION_RULE_LIST(1:3)
+%%%% ¡default!
+Correlation.CORRELATION_RULE_LIST{1}
+
+%%% ¡prop!
+NEGATIVE_WEIGHT_RULE (parameter, option) determines how to deal with negative weights of functional data.
+%%%% ¡settings!
+Correlation.NEGATIVE_WEIGHT_RULE_LIST
+%%%% ¡default!
+Correlation.NEGATIVE_WEIGHT_RULE_LIST{1}
+
+%%% ¡prop!
+DENSITIES (parameter, rvector) is the vector of densities.
+%%%% ¡default!
+0
+
+%% ¡props_update!
+
+%%% ¡prop!
+TEMPLATE (parameter, item) is the analysis template to set the parameters.
+%%%% ¡settings!
+'AnalyzeGroup_CON_FUN_MP_GA_BUD'
+
+%%% ¡prop!
+GR (data, item) is the subject group, which also defines the subject class SubjectCON_FUN_MP.
+%%%% ¡default!
+Group('SUB_CLASS', 'SubjectCON_FUN_MP')
+
+%%% ¡prop!
+G (result, item) is the average multiplex graph obtained from this analysis.
+%%%% ¡settings!
+'MultiplexBUD'
+%%%% ¡default!
+MultiplexBUD()
+%%%% ¡calculate!
+gr = a.get('GR');
+T = a.get('REPETITION');
+fmin = a.get('FREQUENCYRULEMIN');
+fmax = a.get('FREQUENCYRULEMAX');
+densities = a.get('DENSITIES'); % this is a vector
+A = cell(1, 2);
+data = cell(1, 2);
+
+for i = 1:1:gr.get('SUB_DICT').length()
+    sub = gr.get('SUB_DICT').getItem(i);
+    CON_FUN_MP = sub.getr('CON_FUN_MP');
+    
+    % FUN data
+    data_fun = CON_FUN_MP{2};
+    fs = 1 / T;
+    
+    if fmax > fmin && T > 0
+        NFFT = 2 * ceil(size(data_fun, 1) / 2);
+        ft = fft(data_fun, NFFT);  % Fourier transform
+        f = fftshift(fs * abs(-NFFT / 2:NFFT / 2 - 1) / NFFT);  % absolute frequency
+        ft(f < fmin | f > fmax, :) = 0;
+        data_fun = ifft(ft, NFFT);
+    end
+    
+    A_fun = Correlation.getAdjacencyMatrix(data_fun, a.get('CORRELATION_RULE'), a.get('NEGATIVE_WEIGHT_RULE'));
+       
+    % CON data
+    if i == 1
+        data(1) = CON_FUN_MP(1);
+        data(2) = {A_fun};
+    else
+        data(1) = {data{1} + CON_FUN_MP{1}};
+        data(2) = {data{2} + A_fun};
+    end
+    
+    layerlabels = {};
+    for i = 1:length(densities)
+        layerlabels = [...
+            layerlabels, ['C ' num2str(densities(i)) '%'], ...
+            ['F ' num2str(densities(i)) '%']];
+    end   
+end
+
+A(1) = {data{1}/gr.get('SUB_DICT').length()};
+A(2) = {data{2}/gr.get('SUB_DICT').length()};
+
+ba = BrainAtlas();
+if ~isempty(gr) && ~isa(gr, 'NoValue') && gr.get('SUB_DICT').length > 0
+    ba = gr.get('SUB_DICT').getItem(1).get('BA');
+end
+                    
+g = MultiplexBUD( ...
+    'ID', ['g ' gr.get('ID')], ...
+    'B', A, ...
+    'DENSITIES', densities, ... 
+    'LAYERTICKS', densities, ...
+    'LAYERLABELS', cell2str(layerlabels), ...
+    'BAS', ba ...
+    );
+
+value = g;
+
+%% ¡methods!
+function pr = getPPCompareGroup_CPDict(a, varargin) 
+    %GEPPPCOMPAREGROUP_CPDICT returns the comparison plot panel compatible with the analysis.
+    %
+    % PR = GEPPPCOMPAREGROUP_CPDICT(A) returns the comparison plot panel
+    %  that is compatible with the analyze group.
+    %
+    % See also CompareGroup.
+    
+    pr = PPCompareGroup_CPDict_CON_FUN_MP_GA_BUD(varargin{:});
+end
+
+%% ¡tests!
+
+%%% ¡test!
+%%%% ¡name!
+Example
+%%%% ¡code!
+example_CON_FUN_MP_GA_BUD