Author Information

Sanvesh Srivastava
R. W. Doerge

Abstract

Next-generation sequencing technologies have emerged as a promising technology in a variety of fields, including genomics, epigenomics, and transcriptomics. These technologies play an important role in understanding cell organization and functionality. Unlike data from earlier technologies (e.g., microarrays), data from next-generation sequencing technologies are highly replicable with little technical variation. One application of next-generation sequencing technologies is RNA-Sequencing (RNA-Seq). It is used for detecting differential gene expression between different biological conditions. While statistical methods for detecting differential expression in RNA-Seq data exist, one serious limitation to these methods is the absence of biological replication. At present, the high cost of next-generation sequencing technologies imposes a serious restriction on the number of biological replicates. We present a simple parametric hierarchical Bayesian model for detecting differential expression in data from unreplicated RNA-Seq experiments. The model extends naturally to multiple treatment groups and any number of biological replicates. We illustrate the application of this model through simulation studies and compare our approach to existing methods for detecting differential expression such as, Fisher's Exact Test.

Keywords

Hierarchical Bayesian modeling, microarrays, next-generation sequencing, Poisson distribution, differential gene expression, generalized linear models, Gibbs sampling

Creative Commons License


This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Share

COinS
 
May 1st, 12:30 PM

A HIERARCHICAL BAYESIAN APPROACH FOR DETECTING DIFFERENTIAL GENE EXPRESSION IN UNREPLICATED RNA-SEQUENCING DATA

Next-generation sequencing technologies have emerged as a promising technology in a variety of fields, including genomics, epigenomics, and transcriptomics. These technologies play an important role in understanding cell organization and functionality. Unlike data from earlier technologies (e.g., microarrays), data from next-generation sequencing technologies are highly replicable with little technical variation. One application of next-generation sequencing technologies is RNA-Sequencing (RNA-Seq). It is used for detecting differential gene expression between different biological conditions. While statistical methods for detecting differential expression in RNA-Seq data exist, one serious limitation to these methods is the absence of biological replication. At present, the high cost of next-generation sequencing technologies imposes a serious restriction on the number of biological replicates. We present a simple parametric hierarchical Bayesian model for detecting differential expression in data from unreplicated RNA-Seq experiments. The model extends naturally to multiple treatment groups and any number of biological replicates. We illustrate the application of this model through simulation studies and compare our approach to existing methods for detecting differential expression such as, Fisher's Exact Test.