dar: A Consensus-Based Framework for Differential Abundance Testing in Microbiome Data

Abstract

SummaryThe dar R package streamlines differential abundance (DA) testing in microbiome research by integrating state-of-the-art DA methods--such as DESeq2, ALDEx2, ANCOM-BC, and MetagenomeSeq--within a customizable consensus-based framework, thereby enhancing the robustness and reproducibility of DA results. Leveraging microbiome data in phyloseq or TreeSummarizedExperiment formats, dar organizes analysis steps in a modular recipe object, enabling users to easily incorporate preprocessing tasks such as taxonomic filtering, rarefaction, and subsetting, alongside multiple DA analysis methods. Dedicated visualization tools facilitate the definition of a consensus by illustrating the overlap among methods, empowering users to refine analysis strategies and display final results. Reproducibility is supported through functions that export and import entire analysis workflows, making dar a comprehensive solution for addressing the complex, high-dimensional nature of microbiome data. Availability and implementationdar is an R package available from Bioconductor [≥] 3.19 (https://www.bioconductor.org/packages/dar) for R [≥] 4.4. The software is distributed under the MIT License and includes example datasets. Contactfcatala@irsicaixa.es Supplementary informationAdditional documentation is available at https://microbialgenomics-irsicaixaorg.github.io/dar