PCA of methylation profiles

Performs principal component analysis (PCA) on per-sample methylation profiles and plots PC1 vs PC2. Useful for sample-level QC, detecting outliers, and assessing whether biological conditions separate in methylation space.

Usage

plot_pca(object, mod_type = NULL, color_by = "condition", shape_by = NULL)

Arguments

object

A commaData object.

mod_type

Character string specifying a single modification type (e.g., "6mA", "5mC"). If NULL (default), all sites from all modification types are used.

color_by

Character string naming a column in sampleInfo(object) to use for point color. Default "condition".

shape_by

Character string naming a column in sampleInfo(object) to use for point shape. If NULL (default), all points use the same shape.

Value

A ggplot object. PC1 and PC2 are shown on the x- and y-axes, respectively, with percentage of variance explained shown in the axis labels. Each point represents one sample and is labeled with its sample_name. Points are colored by color_by.

Details

Sites with any NA beta values across samples are removed before PCA to ensure a complete data matrix. PCA is computed via stats::prcomp with centering (center = TRUE) and without scaling (scale. = FALSE). A warning is issued if fewer than three samples are present.

See also

methylomeSummary, plot_methylation_distribution

Examples

data(comma_example_data)
plot_pca(comma_example_data)


# Color by condition, shape by replicate
plot_pca(comma_example_data, color_by = "condition")


# Only 6mA sites
plot_pca(comma_example_data, mod_type = "6mA")