JunctionSeq

JunctionSeq is an R/Bioconductor package for detecting differential usage of exons and exon–exon junctions in RNA sequencing (RNA-seq) data. It extends the differential exon usage framework of DEXSeq by incorporating reads that map to exon–junctions, enabling simultaneous testing of changes in exon usage and junction usage within a gene. The method models counts for each exon and junction with a negative binomial generalized linear model, including a gene-level expression component as an offset. It tests whether the relative usage of individual exons or junctions differs between experimental conditions, controlling for overall gene expression and other covariates. The approach yields per-feature statistics and gene-level summaries, allowing assessment of differential transcript usage and alternative splicing events.

Inputs include aligned RNA-seq data (BAM files) and a gene annotation in GTF/GFF format. JunctionSeq requires counts for exons and exon–junctions, which can be generated using its counting workflow or external tools, and a sample table describing conditions and replicates. The package provides functions for model fitting, hypothesis testing, multiple-testing correction, and generation of diagnostic plots and tables. Outputs include lists of exons and junctions with differential usage, adjusted p-values, fold-change estimates, and visualizations that depict usage patterns across samples and conditions.

JunctionSeq is distributed through Bioconductor and is accompanied by documentation and vignettes describing typical workflows, data preparation steps, and interpretation of results. It is commonly used to study differential transcript usage and alternative splicing in RNA-seq experiments, particularly when both exon-level and junction-level signals are informative.