Figure 2

Sources of technical variation for the QC (left) and Investigational (right) data sets prior to the Partek Batch Removal process. A plot of the average F ratio (the signal-to-noise ratio) produced when we applied the ANOVA model (Batch and Spot) to the QC data sets (left), and the Investigational data set (right) which also included ProteinChip lot number (Array). Analysis was performed on spectra after removal of outlier spectra, prior to technical effect batch removal. Batch effect is the largest contributor to variance. The numbers above each group indicate the percentage of peaks that were different in at least one batch (p > 0.005) as determined by the Kruskal-Wallis test. The QC data set is derived from a pooled serum sample so no. After batch removal, none of the peaks were significantly different.