DEXSeq

DEXSeq is a Bioconductor software package for differential exon usage analysis of RNA sequencing (RNA-Seq) data. It is designed to identify exons within genes whose usage differs between conditions, independently of changes in overall gene expression. The method relies on counting reads that map to defined exonic parts, using a transcriptome annotation to partition a gene into non-overlapping exon components.

The statistical approach models exon-level counts with a negative binomial generalized linear model. Counts are analyzed with terms for sample, exon, and gene, along with dispersion parameters. Differential exon usage is assessed by testing the interaction between exon and condition, which probes whether the relative usage of exons within a gene changes across conditions while adjusting for total gene expression.

A typical workflow involves constructing a DEXSeqDataSet from count data and annotation, estimating size factors and dispersion parameters, fitting the GLM, and performing hypothesis tests for differential exon usage. P-values are adjusted for multiple testing to control the false discovery rate.

Outputs include per-exon statistics such as fold changes in exon usage, raw and adjusted p-values, and links to the corresponding gene and exon identifiers. The results can inform interpretation of alternative splicing or exon-level regulatory changes and are often used alongside gene-level differential expression analyses.

Considerations include the reliance on high-quality exon definitions from annotations and potential power limitations with small sample sizes or lowly expressed exons. DEXSeq is widely used in RNA-Seq analysis workflows and integrates with other Bioconductor tools, facilitating downstream analyses and visualization.