Figure 1. Proteins – sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) Broad Range Molecular Weight Standards (Bio-Rad, Hercules, California, USA) – were diluted into four aliquots and applied to 12% Tris–glycine polyacrylamide gels (Novex, San Diego, California, USA) to provide individual protein concentrations of 50, 1.75, 0.0175 and 0.005 ng. Gels were electrophoresed in a 0.1% (w/v) SDS Tris–glycine buffer, pH 8.3, in an X Cell II (Novex) for 1.5 h at 125 V. Gels (a–c) were stained with Colloidal Coomassie Blue (Bio-Rad), SYPRO Ruby Red fluorescent stain (Bio-Rad) and silver. The fluorescent image was captured with a Gel Doc Station (Bio-Rad). The other two stain images were captured with an HP ScanJet 6200 (Hewlett–Packard, Boise, Idaho, USA). Of these detection methods, the fluorescent stain and silver display more sensitivity than the Coomassie stain.

Figure 2. In this set of three hypothetical gels (a–c), spots 1–4 are defined as ‘constitutive’ because they display minimal variation in their density ratios between spots (some of which are illustrated by double-headed arrows). This variation in density ratios is calculated by determining the ratios for each pair of spots within a gel and then identifying those ratios that vary the least across all the gels in the set. Other methods of normalization (such as one that uses the total density of the gel including the background or one that sums the densities of only the spots) could be biased by the occurrence of a dense spot in one gel, such as spot 8 in gel (c), or by density variations of spots with highly variable densities (spots 5, 6 and 7).