PCA plots • ggPopi

ggpop reads PLINK and GCTA PCA output into a typed ggpop_pca object. The same object supports plot_pca() and the layered ggpop() + geom_pca() style.

API summary

Task	API	Notes
Import PLINK/GCTA PCA	`import_pca(file, type = ..., eigenval = NULL, pop_group = NULL)`	Returns `ggpop_pca`
Direct plot	`plot_pca(data, pc_x = 1, pc_y = 2)`	Returns `ggplot`
Layered plot	`ggpop(data) + geom_pca(pc_x = 1, pc_y = 2)`	Tidy ggplot extension path
Compute PCA	`compute_pca(genotype, method = "flashpca")`	Requires `flashpcaR`

GCTA PCA

pca <- import_pca(
  ggpop_extdata("pca", "gcta.eigenvec"),
  type = "gcta",
  eigenval = ggpop_extdata("pca", "gcta.eigenval"),
  pop_group = ggpop_extdata("pop_group.txt")
)

plot_pca(pca)

Scatter chart. Principal component 1 is on the x-axis and principal component 2 is on the y-axis, with point colour encoding imported population groups. Samples cluster by their PCA coordinates while preserving the same population colour mapping used elsewhere in ggpop.

ggpop(pca) + geom_pca(pc_x = 1, pc_y = 3)

Scatter chart. Principal component 1 is on the x-axis and principal component 3 is on the y-axis, showing that geom_pca can switch component axes while keeping the same sample-level PCA data.

When the imported object has a pop column, both plotting routes map population groups to the unified ggpop discrete colour scale.

ggpop(pca) +
  geom_pca(pc_x = 1, pc_y = 2, palette = "population")

Scatter chart. Principal component 1 and principal component 2 are plotted with point colour mapped to population group through the unified ggpop population palette.

PC labels

The axis labels can carry variance explained when eigenvalues are available.

plot_pca(pca, pc_x = 1, pc_y = 2)

Scatter chart. Principal component axes include variance explained percentages in their labels when eigenvalues are available, making the relative contribution of each PC explicit.

Optional flashpcaR computation

compute_pca(method = "flashpca") is available when flashpcaR is installed. It is intentionally not forced during vignette build because it depends on a genotype object.

compute_pca(genotype, method = "flashpca")