Skip to contents

Changelog

Source: NEWS.md

epiRomics 0.99.5

Changes

  • Added public getter and setter methods for all five slots of the epiRomicsS4 class: annotations(), meta(), txdb(), organism(), and genome() (plus the corresponding <- assignment forms). Users should prefer these accessors over obj@slot or methods::slot(obj, "slot"). organism() extends the generic from BiocGenerics and genome() extends the generic from GenomeInfoDb, following Bioconductor convention. Every setter invokes methods::validObject() so invalid assignments (for example an empty-string genome) fail fast with a clear error. A full audit of the codebase confirms these five slots are the only slots on epiRomicsS4; the class definition has not changed, and downstream pipeline functions never introduce new slots — they only update the content of annotations (a GRanges, whose mcols carry the per-stage details).
  • Added ?epiRomicsS4-accessors overview topic that lists every getter/setter pair in one table with runnable examples. The ?epiRomicsS4-class page cross-references each individual accessor via @seealso and documents each slot with an explicit “Access via …” note pointing at the matching getter.
  • Refactored both vignettes to demonstrate the new accessor API. methods::slot() no longer appears in any user-facing example, vignette, or man page. getting-started-with-epiRomics.Rmd had 7 methods::slot() calls replaced with annotations(), plus a prose paragraph noting that users should prefer the getter API (one paragraph inserted in each vignette near first use). articles/full-analysis-with-epiRomics.Rmd had 22 methods::slot() calls replaced.
  • Every roxygen @examples block that previously used methods::slot(obj, "...") now uses the matching getter (touched files: R/synthetic-data.R, R/enhanceosomes.R, R/enhancers.R, R/regions_of_interest.R).
  • Swapped the four runtime methods::slot(database, "...") calls inside make_example_enhanceosome() in R/synthetic-data.R for the matching getters (genome(database), meta(database), txdb(database), organism(database)).
  • Updated user-facing roxygen prose that referred to database@meta or @annotations to reference the new meta() / annotations() accessors instead (touched files: R/synthetic-data.R, R/chromatin_states.R, R/enhancers.R, R/benchmark_enhancer_predictor.R, R/regions_of_interest.R).
  • Added tests/testthat/test-accessors.R with 20 test_that blocks / 50 assertions covering: every getter on the hg38 and mm10 fixtures and on a fresh empty object, every setter round-trip and class preservation, validity enforcement (genome(x) <- "" and txdb(x) <- "" must error), and confirmation that organism() / genome() extend the BiocGenerics / GenomeInfoDb generics rather than shadowing them. After this change the full package test suite runs 261 tests / 607 assertions with 0 failures, 0 errors, 7 legitimate skips (Windows-only harness checks and optional non-model-organism TxDb packages).
  • Added a one-time session tip to the rename-warning helper (.warn_renamed()): the first time a deprecated epiRomics_* alias fires, users also see a one-line note pointing them at the new accessor API (?epiRomicsS4-accessors).
  • Fixed the test-deprecated-aliases.R harness, which was still written against the pre-0.99.4 .Deprecated()-based behaviour and therefore was order-dependent under the message()-based rename helper introduced in 0.99.4. The harness now asserts against the message() condition (via invokeRestart("muffleMessage")) and resets .warn_renamed() state at the top of the block so every alias fires fresh exactly once, regardless of the order other test files ran in. No package code was changed by this test fix.
  • Quality gates for 0.99.5: R CMD check --as-cran reports 0 ERRORs / 0 WARNINGs / 1 NOTE (the expected “New submission” note); BiocCheck reports 0 ERRORs / 0 WARNINGs / 9 NOTES (all pre-existing and stylistic — none introduced by this change); both vignettes rebuild cleanly with the new accessor API.

epiRomics 0.99.4

Changes

  • Replaced .Deprecated() calls in renamed-function wrappers with one-time message() notices via a shared .warn_renamed() helper. Users calling the old epiRomics_* names still see a clear nudge to migrate to the new verb-first names (removal planned for the next release), but BiocCheck’s .Deprecated / .Defunct usage warning no longer fires.

epiRomics 0.99.3

Highlights

  • On-demand example data: cache_data() downloads example datasets (~1.3 GB) via BiocFileCache, keeping the package tarball under 5 MB. has_cache() and get_cache_path() check and locate cached data.

  • Base R graphics track layer: plot_tracks() renders publication-quality multi-track genome browser views in 1-2 seconds per locus using base R graphics. ATAC and RNA signals display in mirrored panels for direct cross-sample comparison.

  • Quick visualization: plot_quick_view() provides zero-setup BigWig signal visualization for any gene locus — no database required.

  • Gene-centered visualization: plot_gene_tracks() displays all database tracks at any gene locus without the enhanceosome pipeline.

  • Chromatin state classification: classify_chromatin_states() classifies regions into biologically meaningful states (active, bivalent, poised, primed, repressed, unmarked) following ChromHMM/Roadmap Epigenomics conventions.

  • TF co-binding analysis: analyze_tf_cobinding() uses Fisher’s exact test with odds ratios and pointwise mutual information (PMI) to identify significant TF co-binding pairs.

API

All exported functions follow Bioconductor naming conventions (verb-first, no package prefix). Legacy names remain exported as .Deprecated() aliases for one release cycle and will be removed in a future version.

Function names:

Parameter names (no deprecation shim — update calls directly):

  • epiRomics_dB → database
  • epiRomics_db_file → db_file
  • epiRomics_genome → genome
  • epiRomics_organism → organism
  • epiRomics_curated_database → curated_database
  • epiRomics_histone, epiRomics_histone_mark_1, epiRomics_histone_mark_2 → histone, histone_mark_1, histone_mark_2
  • epiRomics_track_connection → track_connection
  • epiRomics_index → index
  • epiRomics_keep_epitracks → keep_epitracks
  • epiRomics_type → type
  • epiRomics_test_regions → test_regions
  • epiRomics_putative_enhancers → putative_enhancers
  • epiRomics_putative_enhanceosome → putative_enhanceosome
  • epiRomics_enhanceosome → enhanceosome

The epiRomicsS4 class name is unchanged (the S4 suffix indicates class source).

  • genome is a free-form string validated against the user-supplied TxDb (validate_genome_matches_txdb()). No hardcoded whitelist; any organism with a TxDb.* and org.*.db package is supported.

Dependencies

  • TxDb.Hsapiens.UCSC.hg38.knownGene, TxDb.Mmusculus.UCSC.mm10.knownGene, org.Hs.eg.db, org.Mm.eg.db, BiocFileCache, and parallel moved to Suggests with requireNamespace() guards at every call site.
  • Full @importFrom audit across all R files; NAMESPACE regenerated.

Examples and vignettes

  • Ships a ~0.8 MB toy dataset at inst/extdata/toy/ (400 kb hg38 chr11 window centred on the INS locus) curated from the Zenodo archive (https://zenodo.org/records/19189987). The Getting Started vignette runs end-to-end against the toy data; the full walkthrough is preserved as Full Analysis with epiRomics and resolves the Zenodo archive via cache_data().
  • The full walkthrough lives under vignettes/articles/ as a pkgdown-only article (listed in .Rbuildignore). It is rendered on the pkgdown site but is not built during R CMD check or package installation, so users never wait for the 1.3 GB Zenodo download at install time. The lightweight Getting Started vignette is the only registered vignette in the tarball.
  • Full walkthrough Interactive showcases section links to both the Mouse Islet (Mawla et al. 2023) and Human Islet (Mawla & Huising 2021) Shiny browsers.
  • Exported synthetic-data helpers — make_example_database(), make_example_putative_enhancers(), make_example_enhanceosome(), make_example_bigwig() — power every man-page example so examples run under R CMD check --run-donttest without network access. The only remaining \donttest block is in cache_data() (network download, justified).
  • Both vignettes open with visible setup chunks, interleave explanatory prose before and after every code chunk, and print data-frame structure (manifest CSVs, head() of result objects) so readers see inputs and outputs.
  • Getting-started introduction frames the multi-omics enhancer scope, enumerates Bioconductor integrations, and contrasts epiRomics against Gviz, trackViewer, and Signac.
  • vignettes/references.bib holds package citations; vignettes use [@mawla2023; @mawla2021] inline references.

Documentation and installation

  • README recommends Bioconductor installation first; GitHub install is flagged as not recommended and uses BiocManager::install().
  • README points users to browseVignettes("epiRomics"), the pkgdown site (https://huising-lab.github.io/epiRomics/), and the Bioconductor landing page.
  • pkgdown site restored: _pkgdown.yml and a dedicated .github/workflows/pkgdown.yaml rebuild the site on every push to main and on every release.
  • Top-level doc/ directory removed from the source tree.
  • Comprehensive HTML and PDF vignettes with live BigWig signal visualizations.
  • .onAttach() startup message with version, cache status, and citation info.

Speed

  • Parallel overlap counting via parallel::mclapply() when available.
  • Pre-split annotations by type for O(1) lookup in find_enhanceosomes().
  • Vectorized chromatin state classification and signal aggregation.
  • BigWig RDS caching for memory-efficient large-scale analyses.

CI/CD

  • Cross-platform GitHub Actions matrix: Ubuntu (release/devel/oldrel-1), macOS ARM, macOS Intel, Windows.
  • BiocCheck validation on every push/PR.
  • Test coverage with full integration tests via cached example data.

Housekeeping

  • BiocCheck cleanups: eliminated paste() / paste0() usage inside condition signals across 5 files; removed <<- from R/synthetic-data.R; refactored make_example_enhanceosome() to a synthetic fast path (analyze_tf_cobinding example now runs in <1 s, down from 35 s).
  • Fixed CRAN URL-redirect NOTE by pointing README and vignette links at canonical URLs.
  • Aligned README R-version minimum to DESCRIPTION Depends (R ≥ 4.5.0).
  • Trimmed residual >80-character lines from R source for BiocCheck line-length cleanliness.
  • Dataset provenance wording throughout the vignettes, toy README, and inst/scripts/make-toy-data.R clearly identifies the curator role and points at the package README for the full curation methodology (GEO GSE76268 source; ENCODE-DCC ATAC pipeline; MACS2; DiffBind; FANTOM5, Human Islet Regulome, UCNEs).