Violin plots of the per-cell QC metrics. — plot_metrics

One violin panel per metric, grouped by sample. The metrics are metadata columns, so layer is set explicitly: VlnPlot() otherwise searches for the "data" layer, which does not exist before NormalizeData() has been run.

Usage

plot_metrics_distribution(
  x,
  metrics,
  group_by = "orig.ident",
  color_scheme = NULL,
  file = NULL,
  width = 12,
  height = 6
)

Arguments

x: Seurat object.
metrics: Metadata columns to plot, one panel each. No default - the caller decides, since the right set depends on context (e.g. a single sample can afford pct_ribo/pct_hb alongside detected_genes/total_counts/pct_mito; a multi-sample plot usually can't). A pct_* metric that is under 1% in every cell is skipped (a flat near-zero violin says nothing) - its range is reported with a message instead.
group_by: Metadata column to group the violins by.
color_scheme: (optional) Named vector of colors. Defaults to the "samples" scheme, named by the sorted levels of group_by.
file: Path to save the plot to. NULL returns it without writing. Parent directories are created as needed.
width, height: Saved size in inches. Widen it when there are many samples.

Value

A combined plot, one panel per plotted metric.