Higher-Order Asymptotics for Negative Binomial Regression Inferences from RNA-Sequencing Data

• Activity Number: 571
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Yanming Di*+ and Sarah C Emerson and Daniel W Schafer and Jeff A Kimbrel and Jeff H Chang
• Companies: Oregon State University and Oregon State University and Oregon State University and Oregon State University and Oregon State University
• Keywords: RNA-Seq ; gene expression ; higher-order asymptotics ; negative binomial ; regression

Abstract

RNA sequencing (RNA-Seq) has become the technology of choice for quantifying transcriptome and for studying gene expression. The negative binomial (NB) distribution has been shown to be a useful model for frequencies of mapped RNA-Seq reads. An NB regression model is needed to model gene expression as a function of explanatory variables and to compare groups after accounting for other factors. A considerable obstacle in the development of NB regression is the lack of accurate small-sample inference for the NB regression coefficients. We introduce higher-order asymptotic (HOA) techniques for NB regression inferences. We demonstrate that the HOA-adjusted likelihood ratio tests, although derived from large sample size asymptotic theory, are: 1) essentially exact even for the very small sample sizes typical of RNA-Seq studies, 2) superior to other available tests, and 3) therefore overcome the main obstacle to full development of practical negative binomial regression for RNA-Seq analysis. Additionally, this important application to analysis of biological data will draw attention to HOA, a somewhat neglected yet extremely useful development of modern statistical theory.