%clear all previous variables
clear; format compact


%load the yeast data
load /home/lom/tmp/doublecheck_gsvd/yeast.mat
[nGenes, nExps] = size(data)

%Count and locate NaN Data
nNanPerExp=zeros(nExps,1);
for i=0:nExps-1
  nNanPerExp(i+1) = (sum(sum(isnan(data),2)==i));
end

%Create and display number of missing values
bar(0:nExps-1 ,nNanPerExp, 'r'); axis([-.5 3.5 0 3000])
xlabel('Number of Arrays')
ylabel('Number of Genes')

%Take the only rows without missing values
idx = sum(isnan(data),2)==0;
fullmatrix = data(idx,:);
fullgeneNames = geneNames(idx,:);

%Calculate SVD
[U S V] = svd(fullmatrix, 0);
S=diag(S)'
fractions = S.^2/sum(S.^2);
entropy = -sum(fractions.*log(fractions))/log(nExps)

%plot barchart
subplot(1,3,2) ; barh([1:nExps],fractions, 'r')
title('(b) Expresion Fraction')
axes('position', [.45 .3 .16 .62])
barh([3:12], fractions(3:12), 'r')
set(gca, 'Color', [1 1 .5])

%plot rasterplot
expNames(expNames=='_')=' ';
subplot(1,3,1); rasterplot(V', expNames)
title('(a) Arrays')
colormap(redgreen)
xticklabel_rotate

%plot two first eigengenes
subplot(1,3,3);
plot(V(:,1:2), '-o');
title('(c) Arrays')
ylabel('Expression Level')
set(gca, 'XTick', 1:nExps, 'XTickLabel', expNames)
xticklabel_rotate