Loading plots on a 2D scatter plot — pcaLoading2dPlot • limpca

Produces 2D loading plots from pcaBySvd with the same graphical options as plotScatter as this is a wrapper of this function.

Usage

pcaLoading2dPlot(
  resPcaBySvd,
  axes = c(1, 2),
  title = "PCA loading plot",
  addRownames = FALSE,
  pl_n = 10,
  metadata = NULL,
  drawOrigin = TRUE,
  ...
)

Arguments

resPcaBySvd: A list corresponding to the output value of pcaBySvd.
axes: A numerical vector of length 2 with the Principal Components axes to be drawn.
title: Plot title.
addRownames: Boolean indicating if the labels should be plotted. By default, uses the row names of the loadings matrix but it can be manually specified with the points_labs argument from plotScatter.
pl_n: The number of labels that should be plotted, based on a distance measure (see Details).
metadata: A $n \times k$ "freely encoded" data.frame corresponding to the design argument in plotScatter.
drawOrigin: if TRUE, draws horizontal and vertical intercepts at (0,0) based on the plotScatter function.
...: Additional arguments to be passed to plotScatter.

Value

A ggplot2 object with the PCA loading plot.

Details

pcaLoading2dPlot is a wrapper of plotScatter. See ?plotScatter for more information on the additional arguments.

The distance measure $d$ that is used to rank the variables is based on the following formula: $$d = \sqrt(P_{ab}^2*\lambda_{ab}^2)$$ where $a$ and $b$ are two selected Principal Components, $P_{ab}$ represents their loadings and $\lambda_{ab}$ their singular values.

Examples


data("UCH")
ResPCA <- pcaBySvd(UCH$outcomes)

pcaLoading2dPlot(
  resPcaBySvd = ResPCA, axes = c(1, 2),
  title = "PCA loading plot UCH"
)


# adding color,  shape and labels to points
id_cit <- seq(446, 459)
id_hip <- c(seq(126, 156), seq(362, 375))
peaks <- rep("other", ncol(UCH$outcomes))
peaks[id_hip] <- "hip"
peaks[id_cit] <- "cit"
metadata <- data.frame(peaks)

pcaLoading2dPlot(
  resPcaBySvd = ResPCA, axes = c(1, 2),
  title = "PCA loading plot UCH", metadata = metadata,
  color = "peaks", shape = "peaks", addRownames = TRUE
)
#> Warning: at least one variable used as color or shape is not a factor and will be converted as such


# changing max.overlaps of ggrepel
options(ggrepel.max.overlaps = 30)
pcaLoading2dPlot(
  resPcaBySvd = ResPCA, axes = c(1, 2),
  title = "PCA loading plot UCH", metadata = metadata,
  color = "peaks", shape = "peaks", addRownames = TRUE,
  pl_n = 35
)
#> Warning: at least one variable used as color or shape is not a factor and will be converted as such