This sensitivity and specificity calculation is a nice and over-simplistic way of analyzing two different groups.
There is little and no predictive value build into this software (for this you can use the Matlab classify), because it does not split the data in training or test set, or better in training, test and validation set.

% Evaluates sensitivity, specificity and calculates the
% performance of two different classes.
% Each significant event was analysed for its sensitivity and specificity at different RFU cut off points. The top part of the graph shows a histogram of the normalized RFU values. The case and control RFU values are displayed in different colours and in an ideal world these two histograms show little overlap. A vertical blue line indicates where the best separation/classification is achieved. This scan was performed using all available data rather than splitting the data in test and training sets which would provide a potentially lower sensitivity and specificity.
% The results for each scan are shown in individual plots. The RFU cut off scans display the sensitivity in blue and the specificity in red. The sum of both scans is shown in the green S+S curve above these both lines. The optimal performance point for is reached when the sum of sensitivity and specificity peaks. This position is labelled with a blue dot on the green line.
%
% There are four possibilities for the classification:
%
% True positive = case sample identified as case
% False positive= control sample identified case
% True negative= control sample identified as control
% False negative= case sample identified as control