Skip to contents

Plot ATAC reduction-depth correlations

Source: R/ATACDepthCorPlot.R
ATACDepthCorPlot.Rd

Compute and plot correlations between peak-count totals (or explicit depth metadata) and reduced components, aligning cells by name.

Usage

ATACDepthCorPlot(
  object,
  assay = NULL,
  reduction = "lsi",
  n = 10L,
  depth.column = NULL,
  method = "pearson",
  cells = NULL,
  return.data = FALSE
)

Arguments

object

A Seurat object containing a ChromatinAssay.

assay

Name of the ChromatinAssay; defaults to the active assay.

reduction

Name of a dimensional reduction, normally "lsi".

n

Maximum number of leading components; NULL uses all components.

depth.column

Optional numeric non-negative metadata column. NULL recomputes totals from the selected assay's counts layer, avoiding stale nCount metadata. Peak-matrix totals are not necessarily total fragments or sequencing reads; select a verified fragment-count column explicitly when that is the desired depth measure.

method

Correlation method: "pearson", "spearman", or "kendall".

cells

Cell names in the selected assay; NULL uses all assay cells.

return.data

Return the correlation table instead of a plot.

Value

A ggplot, or a data.frame with component, correlation and n.cells when return.data is TRUE. Constant components have NA correlations. Nothing is written into the object and no component is automatically excluded.

Details

All selected assay cells must occur in the reduction and have finite depth and embedding values; missing cells are not silently dropped. Depth must vary across at least three cells. Component numbers are signed correlations; SVD signs are arbitrary. This diagnostic supports, but does not by itself determine, LSI selection.

See also

RunATACQC()

Examples

if (FALSE) { # \dontrun{
ATACDepthCorPlot(object, assay = "peaks", reduction = "lsi")
ATACDepthCorPlot(object, assay = "peaks", return.data = TRUE)
} # }