Figure 6

The accuracy of peak alignment as a function of gap penalty. The accuracy of dynamic programming peak alignment as a function of gap penalty G, shown on the x-axis. Eight replicate experiments of L mexicana polar extracts were processed, generating eight peak lists. The peak lists were aligned with the dynamic programming for a range of the gap penalty parameter between 0.10 and 0.55 (the retention time tolerance was fixed D = 2.5 s), and the resulting alignment tables were compared to the correct alignment table built manually. The top panel shows the total number of errors in the alignment. The middle panel shows the number of errors of type A (solid line) and type B (dashed line). The bottom panel shows the total number of metabolites in the resulting alignment. The correct number of metabolites is 173, shown in the dashed line.