Gaining Weights ... and Feeling Good about It!

Abstract

Two problems that dog current microarrays analyses are (i) the relatively arbitrary nature of data preprocessing and (ii) the inability to incorporate spot quality information in inference except by all-or-nothing spot filtering. In this chapter we propose an approach based on using weights to overcome these two problems. The first approach uses weighted p-values to make inference robust to normalization and the second approach uses weighted spot intensity values to improve inference without any filtering.