A powerful differential expression test for probe-level oligonucleotide microarray data

​The most common objective of microarray studies is to look for changes in the expression levels of genes, typically comparing data from two biologically distinct sources such as healthy vs. diseased tissue samples. Since microarray experiments are expensive, results from these studies commonly suffer from low numbers of replicates, rendering low statistical power. Here, we propose an alternative analysis based on a paired test of oligonucleotide probe intensities. Using one separate array for characterizing hybridization noise, we develop a Bayesian statistical test for differential expression. We demonstrate that our approach provides low error rates while requiring only a single array hybridization for each biological sample, making gene expression experiments more affordable. We also show that our statistical test predicts error rates with good precision, allowing the researcher to select gene sets in a more rational way.

A powerful differential expression test for probe-level oligonucleotide microarray data.pdf