DEXSeq is a statistical method for detecting differential exon usage in RNA-seq data. It uses generalized linear models and accounts for biological variation to provide reliable false discovery control. DEXSeq detects genes and exons with differential usage and is versatile, applicable to various datasets. It facilitates genome-wide studies of alternative exon usage regulation. DEXSeq is implemented as an R/Bioconductor package. In eukaryotes, a single gene can produce multiple transcripts through alternative splicing. RNA-seq is a powerful tool for studying alternative isoform regulation. However, most methods for detecting differential isoform expression have not accounted for biological variability, leading to unreliable results. DEXSeq addresses this by using a negative binomial distribution and generalized linear models to estimate exon usage and account for biological variation. DEXSeq uses a "flattening" approach to gene models, dividing exons into counting bins. It then models read counts using generalized linear models, incorporating dispersion parameters to account for variability. The method uses a Cox-Reid dispersion estimator to improve accuracy. DEXSeq also allows for information sharing across genes to improve dispersion estimation. DEXSeq was tested on datasets from Brooks et al., Brawand et al., and the ENCODE Project Consortium. It detected significant differential exon usage in many genes, with results showing that DEXSeq outperformed Cuffdiff in terms of false discovery rate control. DEXSeq also provided insights into gene function and regulation by identifying tissue-specific exon usage patterns. DEXSeq is implemented as an R/Bioconductor package and includes functions for data preparation, analysis, and visualization. It can be used on MacOS, Linux, and Windows. The package provides tools for analyzing RNA-seq data, including differential exon usage detection, and allows for downstream analysis using other R or Bioconductor functions. DEXSeq also generates HTML pages with test results and plots, facilitating data sharing and interactive exploration.DEXSeq is a statistical method for detecting differential exon usage in RNA-seq data. It uses generalized linear models and accounts for biological variation to provide reliable false discovery control. DEXSeq detects genes and exons with differential usage and is versatile, applicable to various datasets. It facilitates genome-wide studies of alternative exon usage regulation. DEXSeq is implemented as an R/Bioconductor package. In eukaryotes, a single gene can produce multiple transcripts through alternative splicing. RNA-seq is a powerful tool for studying alternative isoform regulation. However, most methods for detecting differential isoform expression have not accounted for biological variability, leading to unreliable results. DEXSeq addresses this by using a negative binomial distribution and generalized linear models to estimate exon usage and account for biological variation. DEXSeq uses a "flattening" approach to gene models, dividing exons into counting bins. It then models read counts using generalized linear models, incorporating dispersion parameters to account for variability. The method uses a Cox-Reid dispersion estimator to improve accuracy. DEXSeq also allows for information sharing across genes to improve dispersion estimation. DEXSeq was tested on datasets from Brooks et al., Brawand et al., and the ENCODE Project Consortium. It detected significant differential exon usage in many genes, with results showing that DEXSeq outperformed Cuffdiff in terms of false discovery rate control. DEXSeq also provided insights into gene function and regulation by identifying tissue-specific exon usage patterns. DEXSeq is implemented as an R/Bioconductor package and includes functions for data preparation, analysis, and visualization. It can be used on MacOS, Linux, and Windows. The package provides tools for analyzing RNA-seq data, including differential exon usage detection, and allows for downstream analysis using other R or Bioconductor functions. DEXSeq also generates HTML pages with test results and plots, facilitating data sharing and interactive exploration.