Abstract
Protein microarrays consist of known proteins spotted onto solid substrates and are used to perform highly multivariate assessments of protein-binding interactions. Human protein arrays are routinely applied to pathogen detection, immune response biomarker profiling, and antibody specificity profiling. Here, we describe and demonstrate a new data processing procedure, gain-scan, in which data were acquired under multiple photomultiplier tube (PMT) settings, followed by data fitting with a power function model to estimate the incident light signals of the array spots. Data acquisition under multiple PMT settings solves the difficulty of determining the single optimal PMT gain setting and allows us to maximize the detection of low-intensity signals while avoiding the saturation of high-intensity ones at the same time. The gain-scan data acquisition and fitting also significantly lower the variances over the detectable range of signals and improve the linear data normalization. The performance of the proposed procedure was verified by analyzing the profiling data of both the human polyclonal serum samples and the monoclonal antibody samples with both technical replicates and biological replicates. We showed that the multigain power function was an appropriate model for describing data acquired under multiple PMT settings. The gain-scan fitting alone or in combination with the linear normalization could effectively reduce the technical variability of the array data and lead to better sample separability and more sensitive differential analysis.