Skip to content

Plot PCA Rotation

plot_rotation.Rd

Plot the features from a PCA rotation in a 2-dimensional scatter plot. Note: The precedence for the ordering of the coloring is as follows:

  1. col

  2. scores

Usage

plot_rotation(
  data.prcomp,
  dims = 1:2L,
  classes = NULL,
  scores = NULL,
  col = NULL,
  set1 = NULL,
  set2 = NULL,
  set3 = NULL,
  set4 = NULL,
  set5 = NULL,
  lab_cex = 3,
  pt_cex = 2.5,
  auto_ident = TRUE,
  ...
)

Arguments

data.prcomp

A prcomp class object. Typically the object returned by prcomp2().

dims

integer(2). Which dimensions to plot.

classes

Optional. A vector indicating the classes of features used for coloring the points. Must be the same length as the number of features.

scores

Optional. Statistical scores to pass through for the coloring of the points during plotting. If a training data set is passed and scores = NULL, then KS-distances (scores) will be calculated under the hood and used to determine point color. Can be either a single value (e.g. "red") or a vector of color values the same length as the number of observations. This overrides the point color assigned by classes above.

col

character(1). The color of the points. Can be either a single value (e.g. "red") or a vector of color values the same length as the number of observations. This parameter overrides the point colors determined by both the classes and scores parameters above.

set1

Optional. A vector of feature ids to mark on the plot. Typically a vector of analytes of clinical relevance, e.g. sample handling. Marked with a hollow diamond (see pch()).

set2

Optional. An additional vector of feature ids to mark on the plot, separately from those in set1. Marked with a hollow triangle (see pch()).

set3

Optional. An additional vector of feature ids to mark on the plot, separately from those in set1. Marked with a hollow square (see pch()).

set4

Optional. An additional vector of feature ids to mark on the plot, separately from those in set1. Marked with a hollow circle (see pch()).

set5

Optional. An additional vector of feature ids to mark on the plot, separately from those in set1. Marked with a hollow upside-down triangle (see pch()).

lab_cex

numeric(1). Font size of the point labels generated by auto_ident. auto_ident must be TRUE.

pt_cex

numeric(1). Character expansion for the points.

auto_ident

logical(1). Should the top 10 features, in both dimensions, be identified and labeled on the plot?

...

Additional arguments passed to plot_pca_dims().

See also

Other PCA plots: plot_pca_dims(), plot_projection(), screeplot_auc()

Author

Stu Field, Michael R. Mehan

Examples

pca <- center_scale(pcapkg:::log10_ft(simdata), center = TRUE, scale = FALSE) |>
  feature_matrix() |>
  prcomp2()
plot_rotation(pca, col = "green")


class <- withr::with_seed(123, sample(simdata$class_response, 40L))
plot_rotation(pca, classes = class)


y <- withr::with_seed(123, sample(pcapkg:::get_analytes(simdata), 10L))
plot_rotation(pca, set1 = y)