edgeR v4: powerful differential analysis of sequencing data with expanded functionality and improved support for...

Area of Science:

• Bioinformatics
• Computational Biology
• Statistical Genetics

Background:

• edgeR is a widely adopted R/Bioconductor package for differential analysis of sequencing count data.
• It pioneered negative binomial modeling and generalized linear models for complex experimental designs in genomics.
• Empirical Bayes moderation ensures reliable inference even with a small number of biological replicates.

Purpose of the Study:

• To announce and detail the new features and improvements in edgeR version 4.
• To review the statistical framework and computational implementation of edgeR.
• To highlight advancements in differential methylation, transcript expression, and pathway analysis.

Main Methods:

• Utilizes negative binomial distribution and generalized linear models for read count data.
• Implements empirical Bayes moderation for robust statistical inference.
• Introduces infrastructure improvements including C-based model fitting and enhanced quasi-likelihood methods.

Main Results:

• edgeR version 4 incorporates support for fractional counts and improved accuracy for small counts.
• New functionalities include differential methylation analysis, differential transcript and exon usage testing.
• Enhanced capabilities for testing relative to a fold-change threshold and pathway analysis are now available.

Conclusions:

• edgeR version 4 provides significant advancements for differential analyses of various sequencing data types.
• The updated package offers improved accuracy, expanded functionality, and enhanced computational efficiency.
• It continues to be a leading tool for statistical analysis in genomics research.