Principal Components analysis suggests that multiple individualized degradation patterns exist. Panels A and B show a principal components analysis (components 1 and 2) of the entire dataset (61 variables, 10 individuals, 24 and 48 hour time points, expressed relative to t = 0 control sample – ie the data shown in Figure 5). A shaded oval has been added to panel A to show that metabolites from one individual are much more generally sensitive to degradation than those of the others. The shaded ovals in panel B show that this informatics approach readily recognizes the 24 and 48 hour time groups, with the two "exceptions" denoted with arrows. Panel C shows a different principal components analysis (after removal of 24 and 48G from panel A), with the 24 and 48 hour time points from two individuals identified with colored ovals; note that the long axis of the ovals are orthogonal. A similar but visually more difficult relationship to interpret is seen with sample A and F (sample 48F sits just left of the origin at [~-2,0]) Arrows denote two samples in which an individual's samples did not group.
Shurubor et al. BMC Clinical Pathology 2007 7:9 doi:10.1186/1472-6890-7-9