Make a PCA-vs-metadata association heatmap with heatmaply()

Runs anova_pca_metadata on the supplied PCA coordinates and sample metadata, then renders the resulting p value matrix with interactive_heatmap: -log10(p) sets the cell color, the raw p values are shown on hover, and variables (rows) with a single, uninformative value across all shown cells are dropped when cluster_rows is TRUE.

Usage

interactive_pca_metadata_heatmap(
  pca_coords,
  pcameta,
  fraction_explained,
  cluster_rows = TRUE,
  n_components = 10,
  plot_height = NULL,
  ...
)

Arguments

pca_coords

Data frame of PCA coordinates, with samples by row and components by column (e.g. the x element of runPCA's prcomp result).

pcameta

Data frame of sample metadata with sample identifiers by row and variables by column.

fraction_explained

Numeric vector containing the percent contribution to variance of each component (e.g. from calculatePCAFractionExplained).

cluster_rows

Cluster variables (rows) by their p value profile across components?

n_components

Number of leading components to test, passed through to anova_pca_metadata.

plot_height

The total rendered height of the plot in pixels, passed through to interactive_heatmap. Defaults to a height scaled to the number of variables.

...

Additional arguments passed to interactive_heatmap

Value

output A plotly htmlwidget as produced by interactive_heatmap

Examples

pcameta <- data.frame(
  row.names = paste0("sample", 1:6),
  treatment = rep(c("control", "treated"), each = 3),
  batch = rep(c("a", "b"), 3)
)
pca_coords <- matrix(rnorm(6 * 4), nrow = 6, dimnames = list(rownames(pcameta), paste0("PC", 1:4)))

interactive_pca_metadata_heatmap(pca_coords, pcameta, fraction_explained = c(45, 25, 20, 10))