Learn R Programming

cape (version 2.0.2)

plotVariantInfluences: Plot variant-to-variant influences

Description

This function plots the reparameterized influences of variants on each other. The epistatic interactions from the pairwise scan are reparameteriezed to the terms $m_{12}$ and $m_{21}$, where the subscripts indicate the source and target variants respectively. These terms are interpreted as the effect that the source variant exerts on the target variant when both are present. Negative influences represent suppression while positive influences represent enhancement.

Usage

plotVariantInfluences(data.obj, p.or.q = 0.05, min.std.effect = 0, plot.all.vals = FALSE, standardize = TRUE, pos.col = "brown", neg.col = "blue", not.tested.col = "lightgray", light.dark = "light", show.marker.labels = FALSE, show.chr = TRUE, label.chr = TRUE, scale.effects = c("log10", "sqrt", "none"), pheno.width = 11, covar.width = 11, covar.labels = NULL, phenotype.labels = NULL)

Arguments

data.obj
The object in which all results are stored. See read.population.
p.or.q
A threshold indicating the maximum adjusted p value considered significant. If an fdr method has been used to correct for multiple testing, this value specifies the maximum q value considered significant. Only marker pairs with p or q values below this threshold will be plotted.
min.std.effect
A numerical threshold indicating the absolute value of the minimum standard effect size to be shown in the plot. The default value of 0 performs no thresholding.
plot.all.vals
A logical value indicating whether all values should be plotted regardless of significance. If TRUE, the significant values are highlighted in bright colors. If FALSE, non-significant values are not plotted.
neg.col
The color in which to plot the negative values. Colors can be one of the following: "green", "purple", "orange", "blue", "brown", "yellow", or "gray."
pos.col
The color in which to plot the positive values. Colors can be one of the following: "green", "purple", "orange", "blue", "brown", "yellow", or "gray."
standardize
A logical value. If TRUE the values in each entry of the plotted matrix are the standardized effect sizes $\beta/\sigma$. If FALSE, the raw $\beta$ values are plotted.
not.tested.col
A color name used to mark variant pairs that were not tested due to linkage. The color distinguishes these pairs from those that were tested, but do not have significant interactions. The default color is light "gray". This can be changed to "white" or FALSE if no marking is desired.
light.dark
Either "light" or "dark" to indicate what color scheme should be used for plotting.
show.marker.labels
A logical value indicating whether the marker labels should be printed along the plot axes.
show.chr
A logical value. If TRUE, chromosomes are indicated by alternating gray and white blocks along plot axes.
label.chr
A logical value. If TRUE and if show.chr is TRUE, chromosome numbers are printed inside each gray and white block.
scale.effects
A string indicating a scaling function by which the effects should be scaled. This is useful in increasing contrast between effects with large variance.
pheno.width
A numeric value indicating the width of the phenotypes relative to the width of each cell in the interaction matrix. Each cell in the interaction matrix has a width of 1, so a pheno.width of 10 makes the phenotypes 10 times wider for ease of viewing.
covar.width
A numeric value indicating the width of the covariates relative to the width of each cell in the interaction matrix. Each cell in the interaction matrix has a width of 1, so a covar.width of 10 makes the covariates 10 times wider for ease of viewing.
covar.labels
An optional vector of strings specifying a label for each covariate. If this argument is NULL, the covariate names from the data object will be used.
phenotype.labels
An optional vector of strings specifying a label for each phenotype. If this argument is NULL, the phenotype names from the data object will be used.

Value

This function invisibly returns the plotted influence matrix.

See Also

pairscan