Skip to contents

Changelog

Source: NEWS.md

statplot (development version)

Added

  • plot_dot_whiskers: added pvalue_plot_margin argument to parameterize the p-value panel plot margin (points), default c(5.5, 12, 5.5, 0). This prevents x-axis labels on the appended p-value barplot from being clipped when the panel is composed to the right using patchwork.

  • hallmark_pathway_categories: new internal dataset mapping Hallmark pathway names (the term column, with the HALLMARK_ prefix removed) to biological process categories as defined in Table 1 of Liberzon et al. (2015). The dataset includes a runtime validation step against the bundled hallmark_t2g data to ensure pathway names align with MSigDB.

  • plot_covariate_heatmap: new function to draw vertical covariate strips (one strip per covariate). Supports categorical covariates (named color vectors) only.

Changed

  • plot_dotmap: added a short example demonstrating only_show_top_sig (shows top N rows in the combined p-value barplot) and updated the combine_pvalue_method parameter documentation to reference combine_pvalues for method details.

  • combine_pvalues: default methods changed from 'all' to 'CMC'; documentation and tests updated to reflect the new default.

  • plot_dotmap: the default combine_pvalue_method is now "fisher" (was previously "CMC" when not supplied via match.arg()); documentation and examples updated to reflect the new default choice.

  • plot_pvalue_barplot: improved legend behaviour — when a vertical significance line legend is requested (vline = TRUE, vline_legend = TRUE) the colour scale now stores an explicit named values vector so code and tests can programmatically find the named alpha entry; also the fill legend uses override.aes to remove line linetype in legend keys when showing p/q bars so legend keys render as solid/filled swatches.

Fixed

  • plot_dotmap: combine_pvalues() is now called with methods = combine_pvalue_method instead of subsetting the result by name. Previously, combine_pvalues() defaulted to computing only "CMC", so requesting any other method (e.g. "fisher") returned NA, causing the combined p-value barplot to draw no bars.

statplot 0.4.0 - 2026-03-26

Added

  • plot_data_avail_by_group: tile plot showing availability (0/1) of data types across groups, with a black border, no whitespace padding, and optional bold axis labels.
  • plot_pathways: new legend_fixed_dot_size argument (numeric vector) fixes the size-legend breaks and limits to user-supplied gene-count values so that multiple plots composed with patchwork share an identical visual size scale. Values outside the specified range are squished to the nearest extreme rather than dropped.

Changed

  • plot_pathways: the fold_change argument has been renamed to effect_size. Code using fold_change = ... must be updated.
  • plot_pathways: default effect_size_threshold changed from 1.5 to 0 (show all genes by default).
  • plot_pathways: out-of-bounds gene values are now squished (via scales::squish) rather than censored to NA for all color scale paths (diverging, sequential, and in-place colorkey_limits), and for the size scale when legend_fixed_dot_size is set. This prevents genes outside colorkey_limits from rendering as grey.
  • plot_heatmap: when anno_colors is supplied for a categorical covariate, all user-specified color levels now always appear in the annotation legend—including levels absent from the current data (e.g. when plotting a subset). To show only levels present in the data, omit the unwanted levels from the anno_colors named vector.
  • plot_heatmap: added a new merge_legends argument (default FALSE). When TRUE the function:
    • deduplicates annotation legends whose color mappings are semantically equivalent (named vectors are compared after sorting by name; circlize::colorRamp2 functions are compared by their breakpoints and colours), combining their titles (e.g. "pvalue\nqvalue") and suppressing redundant legends;
    • forwards the flag to ComplexHeatmap::draw(..., merge_legends = TRUE) so heatmap and annotation legends are packed together. Unit tests covering the deduplication and semantic comparison behaviour were added.

Fixed

  • plot_dotmap: custom_qvalues passed via ... now correctly drives the q-value bars in the combined p-value barplot. Previously the argument was stripped before being forwarded to plot_pvalue_barplot(), causing the barplot to always display internally computed BH-adjusted q-values instead of the user-supplied ones. The supplied column is now joined from the input data into the combined-p data frame and forwarded as custom_qvalues; when not supplied, the BH fallback is used as before.

statplot 0.3.0 - 2026-03-20

Added

  • run_gsea: wrapper around clusterProfiler::GSEA() for gene set enrichment analysis on a pre-ranked gene vector; returns results alongside the inputs needed by the pathway plotting functions below.
  • plot_pathways: gene–pathway bipartite network (cnetplot) showing the top GSEA pathways and their constituent genes, with nodes colored by fold change.
  • plot_pathway_correlation_network: gene–gene Pearson correlation network for a single pathway, with edges filtered by a correlation threshold and nodes colored by fold change and sized by degree (hub genes appear larger).
  • Example datasets ex_expr_pathway, ex_log2fc_pathway, and hallmark_t2g bundled to support pathway analysis examples and vignettes.
  • plot_numeric_by_3plusgroups for creating violin + boxplots of a numeric variable by a grouping factor, with optional effect size and pairwise Wilcoxon test letters.
  • plot_numeric_by_2groups for creating violin + boxplots of a numeric variable by a grouping factor with exactly 2 levels

Changed

  • plot_pathways: several arguments renamed for clarity:
    • fc_threshold → effect_size_threshold
    • fc_threshold_label → subtitle_effect_size_label
    • show_category → show_pathways
    • size_item → gene_node_size
    • size_edge → line_size
    • category_color / category_size → pathway_color / pathway_label_size
    • item_color / item_size → gene_color / gene_label_size
  • plot_pathways: new max_genes_shown argument adaptively raises effect_size_threshold so that at most max_genes_shown gene nodes are displayed; the chosen cutoff is reported in an auto-generated plot subtitle (subtitle is suppressed when max_genes_shown is NULL).
  • plot_pathways: new subtitle_effect_size_label argument controls the effect-size label shown inside abs() in the subtitle (default "effect size").
  • plot_pathways: new legend_pathway_size_title (default "Num. genes") and legend_color_title (default "Effect size") arguments to customise legend titles for node size and color, respectively.
  • plot_pathways: legend_color_title now accepts plotmath expressions via expression(); validation and documentation updated.
  • plot_pathways: new colorkey_breaks and colorkey_limits arguments to control the tick-mark values and range shown on the color legend.
  • plot_pathways: new color_low, color_mid, and color_high arguments to customise the color scale. Supplying all three produces a 3-color diverging scale; omitting color_mid produces a 2-color sequential scale; leaving all three NULL preserves the default cnetplot palette.
  • plot_pathways: new plot_margin argument (c(top, right, bottom, left) in lines, default c(0.5, 0.5, 0.5, 0.5)) to control whitespace around the plot and prevent node labels from being clipped at the edges.
  • plot_2_categorical_vars: new inside_bar_stats argument controls what is printed inside the stacked bars — "pct" (default, within-group percentage), "n" (count only), "pct_and_n" (percentage and count formatted as pct% (n)), or "none" (no labels).
  • plot_2_categorical_vars: new include_overall_bar argument (default FALSE); when TRUE, a pooled bar showing the marginal distribution of yvar across all observations is prepended to the left of the per-group bars and separated by a vertical line. The bar respects inside_bar_stats and displays the total N label above it like all other bars. The bar label defaults to "Overall" and can be customised via the new overall_label argument.
  • plot_dotmap: new q argument accepts an optional column name of per-cell q-values (e.g. FDR-adjusted p-values) to use for cell shading instead of raw p-values. When both p and q are supplied the combined p-value barplot (right panel) is still computed from p, so Fisher/CMC combination uses raw p-values while cells reflect adjusted values.
  • plot_pvalue_barplot: NA p-values are now handled gracefully — rows with missing p-values are silently dropped rather than causing an error, and BH correction for q-values is applied only to the non-missing subset so adjusted values are not distorted.
  • Use stats::reformulate() with backticks when building formulas in numeric-by-group plotting functions to support non-syntactic column names (fixes issue with spaces/hyphens).

statplot 0.2.0 - 2026-03-06

Added

  • plot_dot_whiskers for plotting effect sizes and confidence intervals, optionally by group and with pvalue barplot
  • plot_heatmap
    • By default, white gridlines around cells
    • Add support for categorical heatmaps
    • Support for clustering categorical data
  • quarto_html_tabset_list: convert a named list object to a tabset when knitting to HTML
  • format_pvalue: function to format pvalues to display in plots
  • combine_pvalues function to combine pvalues to produce a single overall pvalue
    • Updated methods for contemporary best practices
  • plot_pvalue_barplot: horizontal barplot of pvalues
    • By default now shows both pvalues and qvalues stacked
  • plot_dotmap: plot heatmap of effect sizes (dots) with cells shaded by pvalue
    • option to add combined pvalue barplot on the right
      • Changed default combo method to CMC as recommended by literature
    • option to show q-values for the combined pvalue barplot
    • 5 methods are supported when combining pvalues
    • show an x symbol if dot is missing

statplot 0.1.0 - 2026-01-01

Setup first working draft.