Movatterモバイル変換

SUMO 1.2.2

New

• Pretrained artifacts for fast demos/tests Addedcached, pre-trained demo objects (simulated outputs and factorstructures) to speed up examples, vignettes, and CI cold starts.
• Optional backend support: basilisk & reticulateIntroduced backend hooks for isolated Python environments viabasilisk and interoperability viareticulate (aka “reticulare” in earlier notes).Includes auto-detection and clearer setup guidance.
• Demo mode insimulate_twoOmicsData()multiomics_demo = "SHOW" provides a ready-to-run two-factortoy example for quick multi-omics walkthroughs.

Enhancements

• Legacy → standard conversion pipeline Polishedas_multiomics() to reliably upgrade legacysimulate_twoOmicsData() outputs into the current schema(omics, per-omiclist_betas,signal_annotation$features,factor_map, etc.),including:

  • normalization of factor-score names (alpha{n}),
  • mappingdelta{n} →beta{n} for omic2,
  • inference of sample/feature annotations when explicit indices aremissing,
  • robust split ofconcatenated_datasets back intoomic1/omic2 using counts oromic1_/omic2_ prefixes.

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

• Stronger standardization & resilienceAutomatic row/column naming for omics matrices; safer handling ofempty/NA index vectors; improved messages forsample-partition constraints; sturdier concatenated-splitheuristics.

Documentation

• Vignette updates Expanded “Getting Started” andmigration guidance: using demo mode, converting legacy objects withas_multiomics(), and configuring basilisk/reticulatebackends.
• Function docs Roxygen scaffolding added for theinternal utilities and converter while preserving the original functionbodies.

Fixes

• Consistentdelta→beta mapping acrossomics.
• Deterministic naming and factor-label normalization.
• More informative errors when sample segments are too small orfactors are unmapped.

Breaking Changes

• None. Legacy outputs remain supported; callas_multiomics() to standardize them for downstreamworkflows.