Figure 1

Workflow for the optimization of XCMS parameter. A pooled sample is measured sequentially within the studies. The LC-MS data of the pooled sample are then used for optimization. The DoEs are created by using Box-Behnken designs. The individual experiments of the design are calculated in parallel. Peaks are classified as reliable peaks (RP) when they are part of an isotopologue. These RPs serve as basis for the calculation of the Peak Picking Score (PPS). Two additional scores are introduced for retention time correction and grouping. To improve the quality of retention time correction, the relative retention time deviations within the peak groups are minimized which leads to the Retention time Correction Score (RCS). So called ‘reliable groups’ and ‘non-reliable groups’ are defined to assess grouping. The ratio of the squared number of ‘reliable groups’ to ‘non-reliable groups’, the Grouping Score (GS), is maximized within the optimization process. The resulting scores are evaluated by using response-surface-models. The combination of parameters that yields the best score is used as new center for the next DoE. The optimization process continues as long as the respective scores are increasing.