Using credibility intervals instead of hypothesis tests in SAGE analysis

Abstract

Motivation: Statistical methods usually used to perform Serial Analysis of Gene Expression (SAGE) analysis are based on hypothesis testing. They answer the biologist's question: 'what are the genes with differential expression greater than r with P-value smaller than P?'. Another useful and not yet explored question is: 'what is the uncertainty in differential expression ratio of a gene?'. Results: We have used Bayesian model for SAGE differential gene expression ratios as a more informative alternative to hypothesis tests since it provides credibility intervals.