We identified the ProteinChips and fractions with the most informative mass spectra (Table 1), for use in our comparisons. Generally, Fraction 2 (F2) was very sparse in mass peaks on all chips and F5 had a small number of peaks, not consistently detected and generally of a lower quality. Spectra included in analyses had normalization factors
Table 2 presents results comparing the effect of experimental conditions on spectra quality across all batches. These included laser intensity and detector sensitivity adjustments to spot protocols for each fraction, automated applications and defined drying steps. Fewer spectra were excluded in Exp 2 with several ProteinChip/Fraction combinations having no excluded spectra. The largest impact, as assessed by number of spectra removed, appears to be on the CM10 ProteinChip surface (Table 2). Peak selection criteria requested that each peak needed to be present in 80% of spectra to be included in analysis. The number of peaks detected almost doubled in Exp 2 and the CVs for peak intensities improved substantially (Table 2). The quality of the peaks did not improve substantially in term of signal-to-noise ratio (S/N) and resolution, but the intensity of the peaks globally increased, and this contributed probably to better peak detection. The weak cation exchange ProteinChip (CM10), using low stringency (LS) buffers (CM10-LS), under Exp 2 conditions appeared to give the least variant, most complex spectra of all chip surfaces, as determined by number of peaks detected and range of peak intensity CVs (Table 2). We have used peak number as a reflection of spectral complexity. Other parameters can be used to define this, such as S/N ratio, signal height to valley depth.
To determine the reproducibility between the 3 batches in which sera were run we looked for statistical differences in peak intensities between batches using a Kruskal-Wallis non parametric test, adjusted for multiple testing by bootstrapping. We determined the number of peaks in each batch that were statistically different (p 2. Four ProteinChip/Fractions in Exp 2 showed >40% of peaks determined on a single quality control (QC) serum that were statistically different across the batches (Table 2). In general, the CM10-LS fractions showed the lowest batch-to-batch variation.