58
peptides, ‘XRGFRAA’ in muscle and ‘KTXXARX’ in skin where one or more amino-
acids, denoted by an ‘X’, are not identified. We consider the four tripeptides identified
in the fist one and discard counts for the second seven-peptide.
The data reports counts for 4200 tripeptides and 5 tissues over 3 consecutive
stages. For the analysis, we excluded tripeptide-tissue pairs for which the sum of
their counts over the three stages was below 5, leaving n = 2763 distinct pairs. Figure
3.3 shows the parallel coordinates plot of raw data for these tripeptides-tissue pairs.
The desired inference is to identify tripeptide-tissue pairs with an increasing pattern
across the three stages, i.e., to mark lines in the figure that show a clear increasing
trend from first to third stage. Some lines can be clearly classified as increasing,
without reference to any probability model. But for many lines the classification is
not obvious. The purpose of the proposed model-based approach is to define where
to draw the line to define a significant increase.