Reliably determining which genes have a high posterior probability of differential expression: A microarray application of decision-theoretic multiple testing
Abstract
Description
Microarray data are often used to determine which genes are differentially expressed between groups, for example, between treatment and control groups. There are methods of determining which genes have a high probability of differential expression, but those methods depend on the estimation of probability densities. Theoretical results have shown such estimation to be unreliable when high-probability genes are identified.
The genes that are probably differentially expressed can be found using decision theory instead of density estimation. Simulations show that the proposed decision-theoretic method is much more reliable than a density-estimation method. The proposed method is used to determine which genes to consider differentially expressed between patients with different types of cancer.
The proposed method determines which genes have a high probability of differential expression. It can be applied to data sets that have replicate microarrays in each of two or more groups of patients or experiments.
Submitted for publication on 8/14/03
Submitted for publication on 8/14/03