This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| DESeq2 | |
|---|---|
| Name | DESeq2 |
| Developer | Michael Love, Wolfgang Huber, Simon Anders |
| Initial release | 2014 |
| Operating system | Cross-platform |
| Genre | Bioinformatics, Differential expression analysis |
| License | GPL-3.0 |
DESeq2 is a statistical software package for differential expression analysis of count-based high-throughput sequencing data, primarily designed for RNA sequencing. It models count data using the negative binomial distribution and incorporates methods for normalization, variance estimation, and testing for differential expression in complex experimental designs. DESeq2 is widely used in genomics pipelines and has influenced standards in transcriptomics, genome analysis, and computational biology.
DESeq2 was introduced to address reproducible analysis of differential expression in datasets generated by platforms such as Illumina and to integrate with ecosystems including Bioconductor, R Project for Statistical Computing, European Molecular Biology Laboratory, and workflows from groups like ENCODE and The Cancer Genome Atlas. Its statistical framework builds on foundations from earlier tools and concepts advanced at institutions such as Max Planck Society, Wellcome Trust Sanger Institute, and Broad Institute, ensuring compatibility with standards set by consortia like 1000 Genomes Project and Human Cell Atlas.
Development of DESeq2 originated in research labs associated with authors linked to European Molecular Biology Laboratory and Heidelberg University Hospital, following methodological precedents set by packages and studies from groups like Anders and Huber (2010), Robinson and Smyth, and projects at Wellcome Trust Sanger Institute. The package was released via Bioconductor and quickly became part of teaching materials at institutions such as Cold Spring Harbor Laboratory and European Bioinformatics Institute. Adoption accelerated through integration into pipelines used by labs at Stanford University, Harvard Medical School, University of Cambridge, and industry groups at Illumina and GATK-related workflows.
DESeq2 models raw read counts with a generalized linear model based on the negative binomial distribution, employing techniques derived from statistical research performed at places like Johns Hopkins University and University of Oxford. It estimates size factors for normalization leveraging median-of-ratios approaches reminiscent of methods discussed in publications from Broad Institute teams and uses empirical Bayes shrinkage for dispersion and log fold change estimation, an approach connected to work from Bayesians such as Efron and statisticians at Stanford University. Hypothesis testing in DESeq2 uses Wald tests and likelihood ratio tests comparable to methods implemented in packages developed at Harvard University and Imperial College London, accommodating multifactor designs similar to analyses conducted by groups at University of California, Berkeley.
DESeq2 is applied in differential expression studies across biomedical research institutions like National Institutes of Health, European Molecular Biology Laboratory, Wellcome Trust Sanger Institute, and clinical projects within Massachusetts General Hospital and Mayo Clinic. It is used in cancer transcriptomics by researchers associated with The Cancer Genome Atlas, in single-cell pseudo-bulk analyses employed by teams in the Human Cell Atlas, and in microbiome metatranscriptomics work from groups at University of Chicago and ETH Zurich. DESeq2 also features in translational studies by biotechnology companies such as Illumina and Roche and in methodological comparisons within workshops run by Cold Spring Harbor Laboratory and EMBL-EBI.
Benchmarking studies comparing DESeq2 with other tools such as packages developed by researchers at Johns Hopkins University and University of Toronto—including approaches from edgeR, limma, and proprietary pipelines used by Broad Institute—report competitive control of false discovery rate and stable performance across sample sizes common to projects from ENCODE and GTEx. Comparative analyses published by consortia including ENCODE and groups at University of California, San Diego often evaluate sensitivity and specificity against methods introduced by teams at University of Cambridge and King's College London, with DESeq2 frequently favored for balanced trade-offs between power and error control. Performance considerations also reference computational environments provided by Amazon Web Services, Google Cloud Platform, and HPC centers at Oak Ridge National Laboratory.
DESeq2 is implemented in R Project for Statistical Computing and distributed via Bioconductor with vignettes and documentation used in courses at Cold Spring Harbor Laboratory and tutorials at EMBL-EBI. Typical workflows integrate read counting tools and aligners developed by research groups at European Bioinformatics Institute and Broad Institute, such as those from HTSeq, featureCounts, STAR aligner, and HISAT2. Users deploy DESeq2 in pipelines orchestrated by workflow managers developed at institutions like Knime, Nextflow, and Snakemake, often within computational infrastructures provided by XSEDE and cloud services from Amazon Web Services.
Critiques of DESeq2, voiced in comparative methodological papers from researchers at Harvard University, University of Cambridge, and McGill University, note assumptions about the negative binomial model that may not hold in all contexts studied by Human Cell Atlas participants, limitations in handling extreme zero-inflation encountered in single-cell datasets analyzed by groups at Broad Institute, and sensitivity to outliers highlighted by statisticians at Stanford University. Alternative approaches from teams at Wellcome Trust Sanger Institute and EMBL offer complementary models for specific use cases, and ongoing research by institutions such as European Bioinformatics Institute and Max Planck Society continues to refine normalization and shrinkage strategies.
Category:Bioinformatics software