Detecting differential usage of exons from RNA-Seq data

Abstract

RNA-Seq is a powerful tool for the study of alternative splicing and other forms of alternative isoform expression. Understanding the regulation of these processes requires sensitive and specific detection of differential isoform abundance in comparisons between conditions, cell types or tissues. We present DEXSeq, a statistical method to test for differential exon usage in RNA-Seq data. DEXSeq employs generalized linear models and offers reliable control of false discoveries by taking biological variation into account. DEXSeq detect genes, and in many cases specific exons, that are subject to differential exon usage with high sensitivity. We demonstrate the versatility of DEXSeq by applying it to several data sets. The method facilitates the study of regulation and function of alternative exon usage on a genome-wide scale. An implementation of DEXSeq is available as an R/Bioconductor package.

Beyond gene expression level: How are Bayesian methods doing a great job in quantification of isoform diversity and allelic imbalance?

“…To date, it is well known that those features might be associated with the aberrant patterns of disease complexity such as tissue (Anders and Huber, 2010;Anders et al, 2012;Nariai et al, 2014) specific differential expression at isoform levels or tissue specific allelic imbalance in mal-functionality of disease processes, etc. Nevertheless, the knowledge of post-transcriptional modification and AI in transcriptomic and genomic areas has been little known in the traditional platforms due to the limitation of technology and insufficient resolution.…”

“…

AbstractThanks to recent advance of next generation sequencing techniques, RNA-seq enabled to have an unprecedented opportunity to identify transcript variants with isoform diversity and allelic imbalance (Anders et al, 2012) by different transcriptional rates. To date, it is well known that those features might be associated with the aberrant patterns of disease complexity such as tissue (Anders and Huber, 2010;Anders et al, 2012;Nariai et al, 2014) specific differential expression at isoform levels or tissue specific allelic imbalance in mal-functionality of disease processes, etc.

…”

Robust identification of Ptbp1-dependent splicing events by a junction-centric approach in Xenopus laevis

“…We then re-map all the reads in the second pass using both the existing annotation and these newly discovered junctions. In the first approach, we use DEXSeq to identify differential junctions (Diff.junctions), although this package was initially described to infer statistically different exon usage from RNAseq data (Anders et al, 2012). In the second approach, we use the alignment files from STAR with HTSEQ (Anders et al, 2015) and a non-redundant ("flattened" according to HTSEQ terminology) annotation to obtain the number of reads in each exonic regions for each sample, and we again use DEXSeq to identify differential exonic regions (hereafter differential exons (Diff.exon), although they do not necessarily corresponding to bona fide exons due to the flattening).…”

“…using DEXSeq (Anders et al, 2012). The R scripts for differential splicing and follow-up analyses are available on the authors' lab website (https://igdr.univ-rennes1.fr/en/research/research-groups/luc-paillard-group/geneexpression-and-development-group-publications) For the exon-centric analysis, the Xlaevisv1.8.Named.gene.gff3 annotation was made non-redundant ("flattened") using the python script available with DEXSeq (dexseq_prepare_annotation.py) to generate a gtf file composed of non redundant exonic-parts (DEXSEQ.Xlaevisv1.8.Named.gene.exon_reannotated.gtf).…”

“…SpliceTrap generates an exon-trio database (all the possible 3 consecutive exons in the annotation), generates two isoforms for each trio (with or without the middle exon) and quantifies the relative abundances of the two isoforms to infer middle exon usage (Wu et al, 2011). DEXSeq normalizes each individual exon, or non-redundant exonic part, to all the other exons of the gene, and uses generalized linear models to model read counts (Anders et al, 2012). MATS counts the number of reads mapped to the junctions linking each exon to other exons, and the number of skipping junctions, and uses a Bayesian framework to identify differential splicing (Shen et al, 2012).…”