% function [sameSf, diffSf, featNames, sameSfOther, diffSfOther] = ...
%  formNewerFeatureMatrix(dataMat, columnNames)
%formNewerFeatureMatrix - form feature matrices from "newer" data.

function [sameSfPred, sameSfExp, diffSfPred, diffSfExp, featNames] = ...
    predExpFeatureMatrices(dataMat, columnNames)

%% Quick error check
nNames = length(columnNames);
nCols = size(dataMat, 2);
if nNames ~= nCols
  error('Number of column names must be the same as the number of columns');
end


%% First eliminate the data points for which the contact_order is 0
%% as per email exhange with Lars, 5-11 and 5-12 2005.

colIdx = strmatch('prediction_contact_order', columnNames, 'exact');
eliminateIdx = find( dataMat(:,colIdx) == 0);
dataMat(eliminateIdx,:) = [];

colIdx = strmatch('experiment_contact_order', columnNames, 'exact');
eliminateIdx = find( dataMat(:,colIdx) == 0);
dataMat(eliminateIdx,:) = [];


%% Form feature matrices- predicted and exp
featMat = [];
predMat = [];
expMat = [];

%  ... ratio feature ...
colIdx1 = strmatch('prediction_sequence_length', columnNames, 'exact');
colIdx2 = strmatch('experiment_sequence_length', columnNames, 'exact');
featMat = [featMat dataMat(:, colIdx1)./dataMat(:,colIdx2)];
predMat = [predMat dataMat(:, colIdx1)];
expMat  = [expMat dataMat(:, colIdx2)];
%featNames{1} = 'ratio';
featNames{1} = 'length';

%  ... difference in percent alpha ...
colIdx1 = strmatch('prediction_percent_alpha', columnNames, 'exact');
colIdx2 = strmatch('experiment_percent_alpha', columnNames, 'exact');
featMat = [featMat diff(dataMat(:, [colIdx1 colIdx2]), 1, 2)];
predMat = [predMat dataMat(:, colIdx1)];
expMat = [expMat dataMat(:, colIdx2)];
%featNames{2} = 'difference_percent_alpha';
featNames{2} = 'alpha%';

%  ... difference in percent beta ...
colIdx1 = strmatch('prediction_percent_beta', columnNames, 'exact');
colIdx2 = strmatch('experiment_percent_beta', columnNames, 'exact');
featMat = [featMat diff(dataMat(:, [colIdx1 colIdx2]), 1, 2)];
predMat = [predMat dataMat(:, colIdx1)];
expMat = [expMat dataMat(:, colIdx2)];
%featNames{3} = 'difference_percent_beta';
featNames{3} = 'beta%';

%  ... Mammoth zscore feature ...
% copy the same z-score into pred and exp
colIdx = strmatch('zscore', columnNames, 'exact');
featMat = [featMat dataMat(:, colIdx)];
predMat = [predMat dataMat(:, colIdx)];
expMat  = [expMat dataMat(:, colIdx)];
featNames{4} = 'zscore';


% ... experiment contact order (THIS MAY CHANGE) ...
% Why the experimet contact order? It should be the prediction, right?
%colIdx = strmatch('experiment_contact_order', columnNames, 'exact');
%featMat = [featMat dataMat(:, colIdx)];
%featNames{5} = 'experiment_contact_order';
colIdx1 = strmatch('prediction_contact_order', columnNames, 'exact');
colIdx2 = strmatch('experiment_contact_order', columnNames, 'exact');
featMat = [featMat dataMat(:, colIdx1)];
predMat = [predMat dataMat(:, colIdx1)];
expMat = [expMat dataMat(:, colIdx2)];
% featNames{5} = 'prediction_contact_order';
featNames{5} = 'contactOrder';


% additional features not used in classifiers
otherFeatMat = [];
otherFeatCount = 1;
colIdx = strmatch('prob2', columnNames, 'exact');
tmp = dataMat(:, colIdx);
otherFeatMat = [otherFeatMat tmp];
otherFeatNames{otherFeatCount} = 'prob2';
otherFeatCount = otherFeatCount + 1;

colIdx = strmatch('prediction_sequence_length', columnNames, 'exact');
tmp = dataMat(:, colIdx);
otherFeatMat = [otherFeatMat tmp];
otherFeatNames{otherFeatCount} = 'prediction_sequence_length';
otherFeatCount = otherFeatCount + 1;

colIdx = strmatch('convergence', columnNames, 'exact');
tmp = dataMat(:, colIdx);
otherFeatMat = [otherFeatMat tmp];
otherFeatNames{1} = 'convergence';
otherFeatCount = otherFeatCount + 1;

%% Separate same superfamily/different superfamily
colIdx = strmatch('correct_superfamily', columnNames, 'exact');
tmp = dataMat(:, colIdx);
sameSfIdx = find(tmp == 1);
sameSf = featMat(sameSfIdx,:);
sameSfPred = predMat(sameSfIdx, :);
sameSfExp = expMat(sameSfIdx, :);
diffSf = featMat;
diffSfPred = predMat;
diffSfExp = expMat;
clear featMat;
clear predMat;
clear expMat;
diffSf(sameSfIdx,:) = [];
diffSfPred(sameSfIdx,:) = [];
diffSfExp(sameSfIdx,:) = [];

sameSfOther = otherFeatMat(sameSfIdx,:);
diffSfOther = otherFeatMat;
clear otherFeatMat;
diffSfOther(sameSfIdx,:) = [];

return