Pardiff: Inference of differential expression at base-pair level from RNA-Seq experiments

Abstract

In the field of RNA-Seq transcriptomics, detecting differences in expression levels between two data-sets remains a challenging question. Most current methods consider only point estimates of the expression levels, and thus neglect the uncertainty of these estimates. Further, testing for differential expression is often done on predefined regions. Here, we propose Pardiff, a method that reconstructs the profile of differential expression at a base-pair resolution and incorporate uncertainty via the use of a Bayesian framework. This method is built on our approach, Parseq, to infer the transcriptional landscape from RNA-seq data. A program, named Pardiff, implements this strategy and will be made available at: http://www.lgm.upmc.fr/parseq/. © 2013 Springer-Verlag.