Changes in version 1.2.2 (2025-10-14)                  

New

  - Pretrained artifacts for fast demos/tests Added cached, pre-trained
    demo objects (simulated outputs and factor structures) to speed up
    examples, vignettes, and CI cold starts.
  - Optional backend support: basilisk & reticulate Introduced backend
    hooks for isolated Python environments via basilisk and
    interoperability via reticulate (aka “reticulare” in earlier notes).
    Includes auto-detection and clearer setup guidance.
  - Demo mode in simulate_twoOmicsData() multiomics_demo = "SHOW"
    provides a ready-to-run two-factor toy example for quick multi-omics
    walkthroughs.

Enhancements

  - Legacy → standard conversion pipeline Polished as_multiomics() to
    reliably upgrade legacy simulate_twoOmicsData() outputs into the
    current schema (omics, per-omic list_betas,
    signal_annotation$features, factor_map, etc.), including:
    
      - normalization of factor-score names (alpha{n}),
      - mapping delta{n} → beta{n} for omic 2,
      - inference of sample/feature annotations when explicit indices
        are missing,
      - robust split of concatenated_datasets back into omic1/omic2
        using counts or omic1_/omic2_ prefixes.

  - Internal helpers made non-exported (with docs scaffolding) %||% for
    defaults and .is_multiomics_like() for schema checks are now
    internal, with roxygen headers (no code changes).

  - Stronger standardization & resilience Automatic row/column naming
    for omics matrices; safer handling of empty/NA index vectors;
    improved messages for sample-partition constraints; sturdier
    concatenated-split heuristics.

Documentation

  - Vignette updates Expanded “Getting Started” and migration guidance:
    using demo mode, converting legacy objects with as_multiomics(), and
    configuring basilisk/reticulate backends.
  - Function docs Roxygen scaffolding added for the internal utilities
    and converter while preserving the original function bodies.

Fixes

  - Consistent delta→beta mapping across omics.
  - Deterministic naming and factor-label normalization.
  - More informative errors when sample segments are too small or
    factors are unmapped.

Breaking Changes

  - None. Legacy outputs remain supported; call as_multiomics() to
    standardize them for downstream workflows.