Skip to contents

Changelog

Source: NEWS.md

comma 0.6.0

Bioconductor submission preparation

  • Version bumped to 0.6.0 for Bioconductor submission preparation.

  • Package passes R CMD check with zero code-level errors or warnings. Remaining check notes are environment-specific (locale, pdflatex availability) and will not be present in Bioconductor’s check infrastructure.

  • Full roxygen2 documentation for all exported functions, including \donttest{} examples throughout.

comma 0.5.0

Major new features

  • plot_methylation_distribution() — beta value density plot per sample, coloured by sample name, and faceted by modification type when multiple types are present. Useful for QC and comparing methylation level distributions.

  • plot_genome_track() — genome browser-style scatter plot of methylation beta values vs. genomic position. Supports positional windowing via start/ end arguments, mod_type filtering, and optional feature annotation rectangles from annotation(object).

  • plot_metagene() — average methylation profile across a class of genomic features, normalized to fractional position [0 = TSS, 1 = TTS]. Uses annotateSites(type = "metagene") internally.

  • plot_volcano() — volcano plot for differential methylation results. Accepts the data.frame output of results(). Colors points as Hypermethylated, Hypomethylated, or Not significant based on user-supplied padj_threshold and delta_beta_threshold.

  • plot_heatmap() — geom_tile heatmap of the top differentially methylated sites (ranked by adjusted p-value) across all samples. Uses ggplot2 only; no ComplexHeatmap dependency required.

  • plot_pca() — PCA of per-sample methylation profiles using stats::prcomp(). Points colored and optionally shaped by any column in sampleInfo(object).

  • plot_coverage() — histogram of sequencing depth per site, per sample. Useful for coverage QC before differential methylation testing.

Documentation

  • Added two vignettes required for Bioconductor submission:
    • "Getting Started with comma" — end-to-end workflow using comma_example_data.
    • "Working with Multiple Modification Types" — joint 6mA + 5mC analysis.
  • Added package-level documentation (?comma) describing the overall workflow.

Dependency changes

  • Added BiocStyle, ComplexHeatmap, ggrepel, and patchwork to Suggests. None are required for core functions; patchwork enables combined annotation tracks in plot_genome_track() and plot_heatmap().

  • Added VignetteBuilder: knitr to DESCRIPTION.

  • Added Visualization to biocViews.

comma 0.4.0

Major new features

  • diffMethyl() — the core differential methylation function. Accepts a commaData object and a design formula, tests each methylation site for differential methylation between conditions, and returns the object enriched with per-site statistics in rowData. Supports two statistical backends:

    • method = "beta_binomial" (default): per-site quasibinomial GLM via base R stats::glm(). No extra packages required.
    • method = "methylkit": wraps methylKit::calculateDiffMeth() for users who prefer the methylKit pipeline. Requires methylKit to be installed.

    New rowData columns: dm_pvalue, dm_padj (Benjamini-Hochberg by default), dm_delta_beta (effect size: treatment minus control mean beta), and one dm_mean_beta_<condition> column per condition level. Analysis parameters are stored in metadata(object)$diffMethyl_params.

  • results() — S4 method to extract differential methylation results from a commaData object as a tidy data.frame. Supports optional mod_type filtering.

  • filterResults() — S4 method to filter the results table by adjusted p-value (padj) and absolute effect size (delta_beta) thresholds.

  • .parseDorado() — full Dorado BAM parser replacing the previous stub. Reads MM/ML base modification tags from Dorado-aligned BAM files using Rsamtools::scanBam(), parses modification positions from CIGAR-decoded read coordinates, and aggregates per-read calls into per-site beta values. Supports 6mA, 5mC, and 4mC in the same BAM file. Invoked automatically by commaData(..., caller = "dorado").

Dependency changes

  • Added methylKit to Suggests. Required only for diffMethyl(..., method = "methylkit").

comma 0.3.0

Major new features

  • annotateSites() — new vectorized annotation function that replaces annotateMethylSites(), annotateTSS(), and annotateTTS(). Accepts a commaData object and uses GenomicRanges::findOverlaps() internally — no nested for-loops. Supports three annotation modes:

    • type = "overlap": assigns feature type and name to each site; non-overlapping sites are labelled "intergenic".
    • type = "proximity": reports the nearest feature, its distance (in bp), and a signed relative position relative to the feature’s TSS.
    • type = "metagene": reports a fractional position (0–1) within the overlapping feature, strand-aware. Returns an updated commaData with new columns in rowData.

  • slidingWindow() — new generalised sliding window function that replaces methylRollingMedian() and methylRollingMean(). Accepts a commaData object; genome size is always read from genome(object) — no hardcoded values. Supports stat = "median" or stat = "mean", per-sample output, mod_type filtering, and circular genome wrap-around. Returns a tidy data.frame.

  • methylomeSummary() — new function that computes per-sample summary statistics (mean/median/SD of beta values, fraction methylated, mean/median coverage). Returns a tidy data.frame suitable for ggplot2 or tabular reporting. Supports mod_type filtering.

  • coverageDepth() — new function that bins the genome into non-overlapping windows and computes mean or median sequencing depth per window per sample. Replaces calculateMethylSiteDepth(). Returns a tidy data.frame.

  • varianceByDepth() — new function that computes per-sample methylation variance as a function of sequencing depth. Replaces varByCoverage(). Accepts a commaData object; no hardcoded column names. Returns a tidy data.frame.

  • writeBED() — rewritten from scratch. Accepts a commaData object, an output path, and a sample name. Removes all hardcoded developer paths and chromosome names. Writes a standard BED9 file with an optional blue-to-red itemRGB methylation scale.

Breaking changes

  • annotateMethylSites(), annotateTSS(), and annotateTTS() have been removed. Use annotateSites() instead.

  • methylRollingMedian() and methylRollingMean() have been removed. Use slidingWindow() instead.

  • calculateMethylSiteDepth() and varByCoverage() have been removed (they were not exported). Use coverageDepth() and varianceByDepth() instead.

Package changes

  • Root-level clutter removed: functions.R, testscript.R, WT_6mA_Mg.txt, WT_6mA_all_callers.txt, all_site_annotations.txt, and all_site_annotations_60p.txt have been deleted.

  • methylKitGATC.R (historical analysis script) moved to inst/scripts/methylKitGATC_historical.R with a descriptive header comment. It is not part of the package.

  • Version bumped to 0.3.0.

Dependency changes

  • Added GenomeInfoDb to Imports (was transitively available but is now explicitly declared).

comma 0.2.0

Major new features

  • Added commaData S4 class (extends SummarizedExperiment) — the central data object for all comma analyses. Stores methylation beta values and coverage as assay matrices with per-site and per-sample metadata.

  • Added commaData() constructor: parses one or more modkit pileup BED files (or Megalodon files) and merges them into a commaData object with consistent site × sample matrices. Supports multi-sample, multi-modification-type data.

  • Added modkit parser (.parseModkit()): reads modkit pileup BED output (primary ONT methylation calling format). Supports all three modification types: 6mA (a), 5mC (m), and 4mC (21839).

  • Added Megalodon parser (.parseMegalodon()): reads legacy Megalodon per-read BED output for backward compatibility with existing datasets.

  • Added Dorado BAM parser stub (.parseDorado()): not yet implemented; provides a clear error message directing users to use modkit pileup first.

  • Added findMotifSites(): locates all instances of a DNA sequence motif (e.g., GATC, CCWGG) in a genome (BSgenome object or FASTA file) and returns a GRanges of all positions on both strands.

  • Added loadAnnotation(): reads GFF3 or BED annotation files and returns a GRanges with standardized feature_type and name metadata columns.

  • Added accessor functions: methylation(), coverage(), sampleInfo(), siteInfo(), modTypes(), genome(), annotation(), motifSites().

  • Added subsetting: [ for sites/samples and subset() for filtering by mod_type, condition, or chrom.

  • Added internal genome utilities: .validateGenomeInfo(), .circularIndex(), .makeSeqinfo().

Data changes

  • Removed all 9 MG1655-specific bundled datasets (data/*.rda). These were specific to E. coli K-12 MG1655 and not usable with other organisms.

  • Added comma_example_data: a synthetic commaData object with a simulated 100 kb genome, 3 samples (2 conditions), and two modification types (6mA and 5mC). Used in vignettes and tests. Run data-raw/create_example_data.R to regenerate.

  • Added inst/extdata/example_modkit.bed: small modkit pileup BED for constructor tests and examples.

  • Added inst/extdata/example.gff3: matching GFF3 annotation for the example genome.

Dependency changes

  • Added to Imports: GenomicRanges, SummarizedExperiment, IRanges, S4Vectors, BiocGenerics, Rsamtools, zoo

  • Added to Suggests: BSgenome, Biostrings, rtracklayer

  • Removed from Imports: forcats

Package changes

  • Package renamed from CoMMA to comma (Comparative Methylomics for Microbial Analysis). The new name reflects the broader scope — all modification types (6mA, 5mC, 4mC), not just adenine methylation.

  • Version bumped to 0.2.0.

  • R (>= 4.1.0) is now the minimum R version requirement.

  • Added biocViews field in preparation for Bioconductor submission.

Notes

  • The existing functions annotateMethylSites(), annotateTSS(), and methylRollingMedian() are preserved but unchanged. They do not yet accept commaData objects and retain their known limitations (nested for-loops, hardcoded genome size). These will be replaced in Phase 3 (v0.3.0).

comma 0.1.0

  • Initial release as CoMMA (Comparison of Microbial Methylated Adenines).
  • Exported functions: annotateMethylSites(), annotateTSS(), methylRollingMedian().
  • Data: 9 MG1655-specific .rda datasets.