Changes in version 0.99.0                        

  - Initial Bioconductor development release.
  - Introduces the SummarizedExperiment-derived heatmap representation,
    explicit clustering helpers, composable ggplot2 components, and
    patchwork assembly.
  - assay<-/assays<- on a SummarizedHeatmap now invalidate stale
    cluster-derived state: a stored row/column dendrogram is cleared and
    that axis's order reset to natural order, while axes without a
    stored dendrogram (e.g. a manual order) are left untouched. Assay
    replacement never reclusters automatically.
  - SummarizedHeatmap() now preserves the original assay name when
    constructing from a SummarizedExperiment, instead of always renaming
    it to "matrix". An explicit assay argument is now validated even for
    a single-assay SummarizedExperiment, rather than being silently
    ignored.
  - plotHeatmap(collectGuides = FALSE) no longer reserves an empty
    guide-area column: the guide_area() panel and its guideWidth column
    are now omitted from the layout entirely instead of being left
    blank.
  - The SummarizedHeatmap validity method now requires the sole assay to
    be a numeric, non-complex matrix, matching the constructor's own
    check. This means assay<-/assays<- reject replacements that would
    leave a non-numeric, complex, or non-matrix assay.
  - plotRowData()'s and plotColData()'s side argument now names where
    the variable-name axis text is anchored, instead of (for
    plotColData(), a no-op) which side of the heatmap the panel sits on:
    plotRowData(side =) now takes "top"/"bottom" (the narrow axis the
    label sits on) and plotColData(side =) now takes "left"/"right".
    plotHeatmap()'s rowAnnotationSide/colAnnotationSide arguments are
    unchanged and still control panel placement only.
  - The vignette now demonstrates manual column ordering via
    SummarizedHeatmap(colOrder = ).