Quality-control gate diagnostic plot

Plots each unit's statistic against the statistic of the control it is gated on, with the line of equality. Points above the line pass a "greater" gate; points below it fail and are excluded.

Usage

cr_plot_qc_gate(
  gate,
  statistic = NULL,
  reference = NULL,
  fails_median = NULL,
  fails_mean = NULL,
  direction = c("greater", "less"),
  log_scale = TRUE,
  label = NULL,
  title = NULL,
  subtitle = NULL,
  x_lab = NULL,
  y_lab = NULL
)

Arguments

gate

A cr_qc_gate object, a list carrying a units data frame, or a data frame with one row per unit.

statistic

Name of the unit statistic column. NULL (default) auto-detects.

reference

Name of the control statistic column. NULL (default) auto-detects.

fails_median, fails_mean

Names of the logical verdict columns under the two centres. NULL (default) auto-detects; when neither is present the verdict is computed from statistic, reference and direction.

direction

"greater" (default) when a unit must exceed its control to pass, "less" when it must fall below.

log_scale

Draw both axes on a log scale when every value is positive (default TRUE).

label

Optional column whose values annotate the failing units.

title, subtitle, x_lab, y_lab

Plot labels. NULL uses a computed default; NA omits the label.

Value

A ggplot object.

Details

Where the input carries a verdict under both centres, median and mean, the units whose verdict depends on which centre was used are drawn as their own class. That distinction matters because a right-skewed signal has a mean above its median, so gating a raw median against a control mean is silently stricter than the stated rule, and a gate that can only drop low units manufactures apparent effects when it is too strict.

See also

cr_plot_qc() for the distribution dashboard.

Other screen figures: cr_plot_forest(), cr_plot_sample_size(), cr_plot_screen(), cr_plot_specificity(), cr_save_plot()

Examples

set.seed(1)
units <- data.frame(
  well_id = paste0("W", 1:24),
  unit_median = c(rlnorm(20, 5, 0.3), rlnorm(4, 3.9, 0.2)),
  ctrl_median = rlnorm(24, 4.2, 0.1)
)
cr_plot_qc_gate(units, label = "well_id")