Measurement Performance Issues in Microarray Gene Expression Images and Spike-in Experiments

Microarray platforms have been widely used in multiplexed biological measurements. Measurement performance evaluation involves quantitative assessment of several stages of data collection and algorithmic summarization processes, including image analysis, background correction, and statistical modeling. I will review work from NIST Gene Expression Metrology Project in the area of image analysis for microarray gene expression scanner experiments, and multiphase regression statistical modeling for measurement input-output relationships in spike-in microarray experiments. In particular, I will discuss a probabilistic procedure for background correction for data near detection limit and the implications of taking the log-transformation on the data after background correction.