In this repository are the R codes that can be used to perform all the analyses and to create the figures included in the research article entitled "Elementary methods provide more replicable results in microbial differential abundance analysis". The article is published in Briefings in Bioinformatics (Briefings in Bioinformatics, Volume 26, Issue 2, March 2025, bbaf130, https://doi.org/10.1093/bib/bbaf130).
Data_curation folder includes the codes used in the curation of data, e.g. to filter and split datasets (needed in the split-data analyses), and to combine the datasets in one file. Running the script for shotgun datasets (data_curation_shotgun.R) automatically downloads the raw datasets but the raw 16S datasets need to be separately downloaded from https://zenodo.org/records/840333 before running the script data_curation_16s.R. The curated data used in the analyses are, however, also included in the Analysis folder (data_171023.rds, data_meta_041023.rds, data_prevl_061023.rds).
The main results of the paper can be replicated by using the following versions of the R packages: ALDEx2 1.34.0, ANCOMBC 2.8.1 (for ANCOM-BC2), corncob 0.3.2, DESeq2 1.40.2, edgeR 3.42.0, fastANCOM 0.0.4, LDM 6.0.1, limma 3.56.2 (for limma-voom), LinDA 0.1.0, logistf 1.26.0 (for Firth logistic regression), MaAsLin2 1.14.1 (for t-test/linear regression), metagenomeSeq 1.42.0, rms 6.5-0 (for ORM/Wilcoxon), GUniFrac 1.8 (for ZicoSeq)