Skip to contents

dar

Introduction

dar provides recipe-based differential abundance workflows for microbiome data. It accepts phyloseq and TreeSummarizedExperiment inputs and combines preprocessing, quality control and multiple statistical methods through a common, pipeable interface.

A centralized model is required for DA execution. It defines covariates, contrasts, reference levels and the taxonomic level of analysis. Compatibility checks identify which methods can execute that design. Results can be inspected, visualized and compared in a common format while retaining method-specific effect scales. Native options remain explicit, and recipes can be shared through a versioned, non-executable export format.

Consensus is optional: it summarizes agreement across methods for the same contrast and effect direction. Agreement does not guarantee greater accuracy or false-discovery-rate control, and stricter rules can reduce sensitivity.

Installation

You can install the development version of dar from GitHub with:

# install.packages("pak")
pak::pkg_install("MicrobialGenomics-IrsicaixaOrg/dar")

Usage 

library(dar)
data("metaHIV_phy")

## Define recipe
rec <-
  recipe(metaHIV_phy) |>
  add_model(
    ~ RiskGroup2,
    targets = "RiskGroup2",
    tax_level = "Species"
  ) |>
  step_subset_taxa(tax_level = "Kingdom", taxa = c("Bacteria", "Archaea")) |>
  step_filter_taxa(.f = "function(x) sum(x > 0) >= (0.03 * length(x))") |>
  step_maaslin() |>
  step_aldex2()

rec

## Prep recipe
da_results <- prep(rec, parallel = TRUE)
da_results

## Inspect per-method results in a common format
tidy_results(da_results) |>
  dplyr::select(
    taxa_id, comparison, effect_size, effect_metric,
    adj_p_value, significant, method, step_id
  )

The effect_metric column identifies each method’s effect scale; effect sizes from different methods are not necessarily directly comparable. The per-method results can be used without applying a consensus rule.

Optional consensus summary

To add an agreement-based summary, specify the number of supporting methods with bake() and extract the selected taxa with cool():

n_methods <- 2
da_results <- bake(da_results, count_cutoff = n_methods)
da_results

## Results
cool(da_results)

Contributing

Code of Conduct

Please note that the dar project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.