rnb.plot.dreduction: rnb.plot.dreduction

In RnBeads: RnBeads

Description

Creates a dimension reduction plot based on the methylation values of the given dataset.

Usage

rnb.plot.dreduction(
  rnb.set,
  plot.type = "pca",
  dimensions = 1:2,
  distance.metric = "euclidean",
  target = "sites",
  point.types = 0L,
  point.colors = 0L,
  legend.space = 2
)

Arguments

rnb.set	Methylation dataset as an object of type inheriting RnBSet. This dataset must contain at least four samples.
plot.type	Type of plot to be created. This must be one of "pca" (projection to two principal components), "mds" (multidimensional scaling to two dimensions) or "tsne" (t-distributed stochastic neighbor embedding to two dimensions). The section Details provides more details on how the dimension reduction techniques are applied.
dimensions	Vector of two positive integer values giving the principle components to be shown in the horizontal and vertical axis of the plot. This parameter is considered only when plot.type is "pca".
distance.metric	Distance metric to be applied when reducing the dimensionality of the methylation data. This must be one of "eucledian" or "manhattan". The second metric is not supported by principal component analysis.
target	Site or region type to be used in the dimension reduction technique. This must be either "sites" (individual CpGs) or one of the region types summarized in rnb.set.
point.types	Trait, specified as column name or index in the sample annnotation table of rnb.set, to be used to define point types in the plot. Setting this parameter to zero (default) or to a trait that does not define categories results in all samples being displayed as filled circles. If this parameter specifies a column that can be used as sample identifiers, the plot displays the samples as identifiers instead of points.
point.colors	Trait, specified as column name or index in the sample annnotation table of rnb.set, to be used to define sample colors in the plot. Setting this parameter to zero (default) or to a trait that does not define numerical values or categories results in all samples being displayed in black.
legend.space	Width, in inches, of the space dedicated for legends that will be assigned on the right side of the plot. This parameter is considered only if legends are actually included, that is, if sample traits are mapped to point types and/or colors.

Details

The analysis option "exploratory.top.dimensions" controls whether dimension reduction is applied on all probes, sites or regions available in the given dataset, or only on the most variable ones. In case a trait is mapped to point types, the shapes to use are taken from the option "points.category". Similary, the option "colors.category" determines which colors are used when mapping sample categories to color. In cases when numerical values are mapped to color, the option "colors.3.gradient" is used. If the set of value contains both positive and negative numbers, the middle point in the color legend is set to zero. See RnBeads Options for more information on the options mentioned above.

Value

The generated plot as an object of type ggplot. The object also contains an attribute "info", which is a list with the following elements:

"Target"

Targeted sites or regions; the value of the parameter target.

"Technique"

Dimension reduction technique applied; one of "PCA" or "MDS".

"All"

Total number of sites or regions defining the high dimensional methylation space.

"Missing"

Number of dimensions ignored because they contain (only) missing values.

"Selected"

Number of dimensions used when applying a dimension reduction technique.

"Explained"

Value between 0 and 1 showing the variance explained by the selected dimensions, as a fraction of the total variance of all dimensions.

Author(s)

Yassen Assenov

See Also

summarized.regions for listing all region types summarized in a dataset

Examples

library(RnBeads.hg19)
data(small.example.object)
pdf("PCA.pdf", width = 7.2, height = 5.2)
print(rnb.plot.dreduction(rnb.set.example, point.colors="Sample_Group"))
dev.off()

RnBeads documentation built on March 3, 2021, 2 a.m.