Perform PCA on Methylation Data

This function performs Principal Component Analysis (PCA) on methylation (or other) data retrieved from a DuckDB database. It aggregates the chosen value column based on the specified call type and prepares it for PCA analysis.

Usage

plot_mod_pca(
  mod_db,
  call_type = "positions",
  value = m_frac,
  save_path = NULL,
  max_rows = NULL
)

Arguments

mod_db

A list containing the database file path. This should be a valid "mod_db" class object.

call_type

A string representing the name of the table in the database from which to pull the data. Default is "positions".

value

Column to use as the measurement for PCA (e.g., `mh_frac`, `m_frac`). Accepts a bare column name or a single string. Default: `m_frac`.

save_path

Path to save the plot (e.g., .pdf or .png). If NULL, the plot is not saved.

max_rows

Optional maximum number of rows to sample from the table (for speed on large datasets).

Value

Produces a PCA plot (PC1 vs PC2) and prints a PCA summary and the PCA scores.

Details

The function connects to the specified DuckDB database, retrieves data from `call_type`, reshapes to a features × samples matrix (features are regions/windows/positions; columns are samples), scales features, then runs PCA on samples (transpose before `prcomp`).

Examples

if (FALSE) { # \dontrun{
# Default (m_frac)
 plot_mod_pca(mod_db)
 # Use mh_frac instead
 plot_mod_pca(mod_db, call_type = "regions", value = mh_frac)
 # Or as a string
 plot_mod_pca(mod_db, call_type = "windows", value = "mh_frac")
} # }