Analysis of cDNA Microarray Experiment with External Controls : A Mixed Model Approach

Abstract

The aim of this article is to obtain the calibration curves by external controls in the cDNA microarray data set. Anonlinear mixed model is found to be suitable for this purpose. The fitted calibration curves could be used to predict the concentrations of the relative expressions by the inverse functions of the fitted curves. After giving some background introduction for a eDNA microarray experiment with external controls, the article focuses on two parts. First, we check the source of experiment variation by using linear model on the slide, dye, and pin effects, and determine the quality of the data set by using quasi­likelihood method for the model with random effect for pins. Second we propose a non­linear model based on a four­parameter logistic function, with two possible random effects structures on pins and base on the concentrations. The inverse function of the fitted model is then used to predict the concentrations from the expression intensities of the external controls.