Skip to contents

Changelog

Source: NEWS.md

MicrobiomeStat 1.5.1

Bug Fixes

  • Fixed incorrect where() syntax in dplyr::across() calls — where(is.character) & !is.factor is not a valid tidyselect expression; corrected to where(~ is.character(.) & !is.factor(.)) in four alpha diversity files.
  • Fixed .data$ pronoun used in base R subsetting in generate_taxa_cladogram_single.R, which caused significance filtering to silently fail.
  • Fixed interface_utils.R so that generate_taxa_test_pair is correctly recognized as a function requiring change.base.
  • Fixed missing(feature.level) check in linda() to avoid error when the argument is explicitly passed as NULL.
  • Fixed mStat_rarefy_data.R to use rowname matching instead of positional indexing when subsetting feature.ann after rarefaction.
  • Fixed missing drop = FALSE on matrix subsetting in plot_feature_diversity.R and mStat_subset_dist.R that could collapse results to vectors.
  • Fixed mStat_subset_data.R to use short-circuit && instead of vectorized & in a scalar conditional.
  • Fixed mStat_calculate_beta_diversity.R to use warning() instead of message("Warning: ...") for capitalization mismatches.
  • Fixed knitr chunk option typo (result → results) in 9 places across report section generators.
  • Fixed hardcoded pdf = TRUE → pdf = pdf in report section generators to respect the caller’s parameter.
  • Fixed describe_change_method() context from 'alpha' to 'taxa' in the taxa change heatmap report section.
  • Removed duplicate section header in mStat_generate_report_long_sections.R.

Code Quality

  • Replaced single &/| with short-circuit &&/|| in scalar if-conditions across 9 files.
  • Standardized PDF file naming across 12 beta and 3 alpha files to use the centralized mStat_append_pdf_group_suffixes() helper instead of manual paste0 with inline NULL checks.
  • Fixed data.obj <- → analysis_data.obj <- normalization assignment pattern in 6 taxa files that were passing the original data object instead of the normalized one to downstream analysis.
  • Removed dead code: unused variables (midpoint, midpoints, avg_abund, Time_choices, result), a no-op else { p <- p } branch, an unused perplexity parameter, and a dead feature.change.func string conversion.

MicrobiomeStat 1.5.0

New Features

  • Added a unified analysis interface and expanded data handling options, including explicit support for "other" feature data flows and consistent feature-level requirements across analysis paths.
  • Added linda2() with detection-depth weighting and tree-guided smoothing support, and exported it for direct package use.
  • Added ref.level support to taxa differential testing helpers so reference groups can be chosen explicitly instead of relying on factor ordering.

Statistical and Workflow Fixes

  • Fixed time semantics, metadata alignment, and interface contract regressions across single, paired, and longitudinal workflows.
  • Fixed multiple plotting and report-generation regressions so examples, reports, and exported figures remain runnable and consistent with package defaults.
  • Fixed remaining R CMD check and dependency handling issues, including a cleanup of GitHub Actions dependency resolution for package checks.

Internal Improvements

  • Refactored taxa helpers, filter handling, and report-section generators to reduce duplication and make plotting/report code paths more consistent.
  • Centralized documentation templates and regenerated Rd files to keep exported interfaces aligned with implementation changes.

MicrobiomeStat 1.4.4

New Features

  • Added ref.level parameter to differential abundance testing functions for specifying reference group:

  • The ref.level parameter allows users to specify which group level should be used as the reference for comparisons, instead of relying on alphabetical ordering (first level alphabetically). This provides more flexibility and control over the statistical comparisons.

  • Example usage:

    # Set "Control" as reference instead of alphabetically first group

test.list <- generate_taxa_test_single(
  data.obj = data.obj,
  group.var = "treatment",
  ref.level = "Control",  # New parameter
  feature.level = c("Genus"),
  ...
)

MicrobiomeStat 1.4.3

Bug Fixes

  • Fixed variable type handling in generate_taxa_test_single(): The function now correctly distinguishes between categorical (factor/character) and continuous (numeric/integer) variables. Previously, all variables were unconditionally treated as categorical.

Documentation Improvements

  • Updated roxygen documentation for generate_taxa_test_single():
    • Fixed @return description to accurately reflect actual output columns
    • Added comprehensive explanation of variable type handling (categorical vs continuous)
    • Added detailed statistical methods documentation
    • Clarified difference between LinDA (for count/proportion data) and linear models (for other data)

Code Quality

  • Removed legacy code from generate_taxa_volcano_single():
    • Removed unused meta_tab variable extraction
    • Removed unused group_level and reference_level variable definitions
    • Code is now cleaner and more maintainable

Testing

  • Verified continuous variable support in volcano plots
  • All tests pass with both categorical and continuous variables

MicrobiomeStat 1.4.2

Bug Fixes

  • Fixed pseudocount imputation for taxa change analysis
  • Fixed log fold change pseudocount bias in taxa change functions
  • Fixed critical statistical issues in normalization and CLR transformation
  • Fixed barplot/areaplot mean calculation

New Features

  • Added linda2() function with experimental sample weighting support
  • Exported linda2 function for external use

MicrobiomeStat 1.4.1 and earlier

See git commit history for details.