IMPROVED: Full Monte Carlo validation now enforces calibrated LinDA bias and coverage gates, compares results with a reviewed stochastic baseline, records dependency versions and publishes concise execution summaries (#155).
IMPROVED: Full validation reports are retained as compact machine-readable history on the validation-results branch, while weak longitudinal contrasts remain descriptive rather than receiving inappropriate power gates (#155).
DEPRECATED: Legacy selectors and model-free differential-abundance execution remain available in Bioconductor 3.24, become defunct in 3.25 and are scheduled for removal in 3.26 (#157).
CHANGED: Require R 4.6 and make ComplexHeatmap, gplots, heatmaply and UpSetR optional plotting dependencies with classified installation guidance (#157).
IMPROVED: Remove generated test artifacts and address actionable BiocCheck findings in documentation, vignettes and package metadata (#157).
dar 1.9.11
FIXED
Consensus tables, plots and bake results now keep modeled targets, contrasts and effect directions as separate hypotheses instead of mixing their support counts (#154).
Modeled intersections use the complete taxon-by-contrast universe, including non-significant hypotheses, and classify zero effects as neutral (#154).
Method exclusions are applied before consensus support is counted, and invalid cutoffs, weights and exclusions now produce classified errors (#154).
Modeled exploration functions accept explicit target and contrast_id selectors; abundance plots require them whenever automatic selection would otherwise be ambiguous (#154).
dar 1.9.10
BREAKING CHANGES
CHANGED: sample_data() and tax_table() now return the complete recipe metadata and taxonomy instead of model-restricted analysis views (#153).
CHANGED: otu_table() now always returns a wide taxa-by-sample tibble with deterministic taxon and sample ordering, regardless of the stored phyloseq orientation (#153).
IMPROVED: Differential-abundance engines and plots use explicit internal target-only and selected-rank views, preventing public accessor semantics from changing analysis inputs (#153).
dar 1.9.9
BREAKING CHANGES
CHANGED: export_steps() and import_steps() now use a real, versioned, non-executable JSON schema. Files produced by the legacy pseudo-JSON format are rejected with migration guidance instead of being evaluated (#152).
CHANGED: Imported bake configurations require an explicit prepare = TRUE before import_steps() may execute an analysis on an unprepared recipe (#152).
IMPROVED: Imported constructors, arguments, formulas, nested values and filter predicates are allowlisted and validated transactionally before a recipe is changed (#152).
dar 1.9.8
FIXED: Recipe steps are classified by their concrete classes, so arbitrary user-defined IDs no longer change preprocessing, differential-abundance or bake semantics (#151).
FIXED: Preprocessing operations execute in their configured order, including recipes that interleave filtering, subsetting and rarefaction (#151).
IMPROVED: Unsupported, misplaced and duplicate-ID recipe steps now fail with classified validation errors before execution (#151).
dar 1.9.7
BREAKING CHANGES
DEFUNCT: phy_qc() now directs users to recipe_qc() instead of inferring an analysis target implicitly (#112).
NEW: recipe_qc() provides model-independent overall or explicitly grouped quality-control metrics for both Recipe and PrepRecipe objects (#112).
IMPROVED: QC summaries expose unambiguous group_by and group columns and reshape correctly oriented count data only once per call (#112).
dar 1.9.6
NEW: step_linda() integrates bias-corrected LinDA linear and mixed-effects models with every compatible centralized contrast (#128).
IMPROVED: LinDA participates in the canonical tidy result contract and the Monte Carlo scientific-validation workflow with explicit log2 fold-change semantics (#128).
dar 1.9.5
NEW: The seven centralized DA engines accept structured engine_args for advanced arguments at explicit native-call stages (#126).
IMPROVED: Step export/import preserves nested advanced arguments and rejects values that cannot be reconstructed faithfully (#126).
dar 1.9.4
NEW: tidy_results() exposes a stable, engine-labelled result contract while preserving the engine-specific tables stored in PrepRecipe@results (#127).
IMPROVED: Result consumers validate taxon-contrast-step keys and use explicit effect metrics and p-value mappings instead of method-specific aliases (#127).
dar 1.9.3
BREAKING CHANGES
CHANGED: add_model() is now the single source of truth for differential abundance analysis targets and taxonomic resolution through its new targets and tax_level arguments (#142).
DEPRECATED: add_var(), add_tax(), get_var(), get_tax(), the var_info and tax_info arguments to recipe(), and model-free DA execution now emit classified migration warnings. They remain functional during this first Bioconductor deprecation cycle (#142).
IMPROVED: Legacy centralized models, serialized recipes and imported step files are normalized to the new model representation without changing the original microbiome metadata (#142).
FIXED: Modeled exclusion and mutual plots preserve contrast and effect-direction keys, normalize adjusted p-value aliases without duplicate columns, and LEfSe executes every planned pairwise contrast for multilevel targets (#142).
dar 1.9.2
NEW: Add a centralized statistical model with add_model() and get_model(), including fixed confounders, target-time interactions, supported random effects, explicit references, and a shared missing-value policy (#130).
NEW: Generate deterministic condition-within-time, time-within-condition and optional difference-in-differences contrasts, with a common result contract across DESeq2, ALDEx2, ANCOM-BC2, corncob, MaAsLin3, Wilcoxon and LEfSe (#130).
IMPROVED: Record executed and skipped methods in prepared recipes, validate engine capabilities before execution, and keep consensus results separated by contrast and effect direction (#130).
IMPROVED: Preserve centralized model specifications when exporting and importing recipe steps (#130).
dar 1.9.1
IMPROVED: Enforce structural invariants for Recipe and inherited PrepRecipe objects through S4 validity checks (#124).
IMPROVED: Validate add_var(), add_tax(), and preprocessing results early, with contextual errors that identify the failing step (#124).
dar 1.5.6
CHANGED: Mark step_metagenomeseq as defunct and remove its implementation and related references.
dar 1.5.5
CHANGED: Migration from Maaslin2 to maaslin3
dar 1.5.4
FIXED: Resolved a crash occurring on servers with >128 cores by limiting connection setup based on available R connections.
CHANGED: Refactored parallel execution to use with(..., local = TRUE). This ensures the internal parallel plan is temporary and does not overwrite the user’s global future configuration.
Fixing bug in Github Actions on Linux with rlang installation.
dar 0.99.3
New Features
The dar package now accepts both phyloseq class objects and TreeSummarizedExperiment as inputs.
The tutorial has been refocused to become a tutorial on how to import biom, qiime, mothur, metaphlan, and humann into TreeSummarizedExperiment and phyloseq class objects.
The Recipe and PrepRecipe classes have been introduced, replacing the previous recipe and prep_recipe classes.
The subset and filter operations have been updated to allow all steps of the recipe to be defined in a chainable manner.
The functions step_filter_by_abundance, step_filter_by_prevalence, step_filter_by_rarity, and step_filter_by_variance have been added to enhance filtering functionality.
Improvements
The R version dependency has been updated to 4.4.0.
The dependency on data.table has been removed.
The re-export of %>% and := has been removed. Now code examples and vignettes use |>.
A warning message is now displayed whenever the rarefy = TRUE option is used, informing users that a fixed seed is being used and how it could impact their results.
The package coverage has increased to 82.33%.
Bug Fixes
Unconventional package installation methods have been avoided, for example, pak::pkg_install.
The setting of a seed within a function (run_aldex) has been addressed.
The name of the data in the R/data.R documentation has been corrected from NA.
