Plot a consensus rank-selection (nmfkc.consensus) object

Two views of a nmfkc.consensus result:

• type = "criteria" (default): the stability curves cophenetic (blue) and dispersion (red) against rank. No "best" marker is drawn – a stable rank tends to be a coarse one, so the curves are shown for inspection rather than as an optimum.

• type = "heatmap": the consensus matrix of each requested rank, reordered by average-linkage hierarchical clustering of \(1 - \bar C\) (blue = 0 / different cluster, red = 1 / same cluster). Requires keep.consensus = TRUE at compute time.

Usage

# S3 method for class 'nmfkc.consensus'
plot(
  x,
  type = c("criteria", "heatmap"),
  rank = NULL,
  mfrow = NULL,
  mar = NULL,
  col = grDevices::hcl.colors(50, "Blue-Red"),
  main = NULL,
  ...
)

Arguments

x

An object of class "nmfkc.consensus".

type

"criteria" or "heatmap".

rank

For type = "heatmap", the rank(s) to display. NULL (default) shows every rank stored in x.

mfrow

Panel layout for multiple heatmaps, as c(nrow, ncol). NULL (default) uses n2mfrow for a near-square grid.

mar

Per-panel margins c(b, l, t, r) for the heatmap view. NULL (default) uses tight margins for a grid and wider margins (for sample labels) when a single rank is shown.

col

Heatmap colour palette (length-50 blue-to-red by default).

main

Plot title for the "criteria" view (heatmap panels are titled per rank).

...

Further arguments passed to the underlying plot / image.

Value

x, invisibly.

See also

nmfkc.consensus