ruv_svdplot {ruv}

RUV SVD Plot

Description

A generalization of a PC (principal component) plot.

Usage

ruv_svdplot(Y.data, Y.space = NULL, info = NULL, k = c(1, 2), Z = 1, left = TRUE)

Arguments

Y.data

The data matrix. Rows are observations and columns are features (e.g. genes).

Y.space

Either a data matrix of the same dimension as Y.data, or the SVD of such a matrix, as returned by the svd function. The singular vectors of this matrix define the space in which Y.data will be plotted. If NULL, Y.data itself is used.

info

Additional data to be included in the ggplot, which can be used for setting aesthetics such as color. Converted to a dataframe, which should have a number of rows equal to the number of rows of Y.data (if left=TRUE) or the number of columns of Y.data (if left=FALSE).

k

A numeric vector of length 2. The singular vectors to be plotted. Typically integers, but fractional values can also be specified. For example, a value of 2.5 corresponds to the linear combination (singular vector 2) + (singular vector 3), rescaled to have unit length. Similarly, a value of 2.2 corresponds to the (rescaled) linear combination 8*(singular vector 2) + 2*(singular vector 3), and -2.2 corresponds to the (rescaled) linear combination 8*(singular vector 2) - 2*(singular vector 3). Note that the vectors defined by 2.2 and -2.8 are orthogonal to each other, as are those defined by 2.3 and -2.7, etc.

Z

Any variables to regress out of Y.data as a preprocessing step. May simply be 1 (the default) for an intercept term, i.e. the columns of Y are mean centered. May also be NULL. Similarly for Y.space, unless Y.space is already an SVD.

left

Plot the left singular vectors (if TRUE) or the right singular vectors (if FALSE).

Details

When Y.space = NULL and Z = 1 and the values of k are integers, this is a standard PC plot.

Value

A ggplot.

Author(s)

Johann Gagnon-Bartsch