Changes in version 0.99.4                        

  - Declared R.utils in Suggests: data.table::fread() needs it to read
    the gzipped demo matrix that the vignette and read_input_matrix()
    examples use. Its absence broke the 0.99.3 vignette build on the
    Bioconductor build machines (it was present locally, which masked
    the missing declaration).

                       Changes in version 0.99.3                        

Changes responding to the first round of Bioconductor review:

  - path2path() now defaults to verbose = FALSE; build_null_pool() and
    calibrate_pvalues() gained a verbose argument (default FALSE) gating
    the per-permutation progress messages, which were previously
    unconditional. Messages that signal a change of behavior (GPD
    fit-adequacy fallback, collection-name aliasing, non-default
    configuration reuse, names appended beyond a user-supplied plot
    order) remain.
  - All imported functions are now declared with importFrom() in
    NAMESPACE.
  - Function documentation states the class of every parameter.
  - The vignette gained a section on how the package integrates with the
    Bioconductor ecosystem and how it differs from per-contrast
    enrichment packages; the quick-tour chunks explain each parameter
    and interpret the printed output; and all code chunks are now
    evaluated, including the Route B calibration (run at B = 50 on the
    demo data) and the file-based input example, which reads the demo
    matrix shipped in inst/extdata/zscore_demo.csv.gz.
  - README rewritten around the user-facing question the package
    answers, with a copy-paste quick start that runs on the bundled demo
    data.
  - Added the WholeGenome biocView.

                       Changes in version 0.99.2                        

  - Replaced repeated per-ranking fgsea tied-statistics warnings with
    one summary per stage. The summary reports the number of affected
    rankings and their median and maximum tie fractions; other warnings
    remain visible.
  - Applied the same tie reporting to the pooled implementations and
    added regression tests across Stage 1 and Stage 2 modes.

                       Changes in version 0.99.1                        

  - Added conversion to and from SummarizedExperiment.
  - Strengthened input validation, reproducibility, and documentation in
    preparation for Bioconductor review.
  - When no adequate GPD tail fit can be obtained, calibrate_pvalues()
    returns pooled empirical p-values instead of extrapolating from an
    inadequate fit.

                       Changes in version 0.99.0                        

  - Initial Bioconductor submission candidate, implementing two-stage
    pathway mapping, two significance routes, overlap masking, plotting,
    and file input/output.