3.5. Data Analysis
So far, MicroVigene array software (version 5.6, VigeneTech Inc., Carlisle, MA, USA) is the only software—at least to our knowledge—that was initially designed for RPPA data analysis that includes automatic spot finding and background subtraction. However, beside MicroVigene other programs are also appropriate for RPPA data analysis. These include Array Pro (version 6.3, Media Cybernetics, Rockville, MD, USA), GenePix Pro (version 7.2.29, Molecular Devices, Sunnyvale, CA, USA), and Mapix (version 7.3.1, Innopsys, Carbonne, France) [10]. The spotting of dilution series allows quantification of the protein levels of each sample. To obtain the exact protein concentration, values representing the dilution series should be in a linear range. Positive and negative controls are recommended for use in certification of the measured spot signal. Problems that might occur during data analysis are variations of the background intensities, emerging due to non-specific binding of the secondary antibody or uneven exposure of different parts of a RPPA slide to the reagents used in protein detection (e.g., ECL solution). A recently published study addresses this problem of spatial heterogeneity by the use of positive controls in duplicates. Depending on their spot intensity spatial variation could be corrected for each slide [67]. In addition, Neeley et al. propose a normalization step which mainly removes inter-array variability [68]. Another possibility is the spotting of a cell line panel on each slide to be able to calculate spot sizes of the protein of interest normalized to the average spot intensity of the cell lines. In a next step, the calculated data have to be normalized in order to correct potential sources of variability that do not reflect biological differences in protein levels between the investigated samples, such as variations of total amount of protein extracts that were spotted. The usual practice of normalization to a housekeeping (endogenous reference) protein can be problematic due to the limitations of “true” housekeeping proteins. A common method to normalize RPPA data are dye binding methods, i.e., measurements of total protein on one slide per print run with dyes like Fast Green FCF, Sypro Ruby, or colloidal Gold. The signals obtained reflect the total amount of protein per spot immobilized on the solid phase [69,70]. An alternative approach has been shown in a study where an antibody against single-stranded DNA (ssDNA) can be used as a suitable RPPA normalization parameter [71].