Population frequencies with the uncertainty of the gate that produced them

One row per population, one point per sample, and a bar through each point spanning plus and minus the standard uncertainty of that sample's frequency propagated from where its thresholds were placed.

Usage

fig_frequency_uncertainty(
  ufreq,
  outfile,
  group_of = NULL,
  dpi = 200,
  colors = fcs_colors()
)

Arguments

ufreq

the frequencies element of run_gate_uncertainty()

outfile

Path to write the figure to.

group_of

Named character vector mapping sample_id to group label.

dpi

Resolution in dots per inch. Default 200.

colors

Named list of colours; defaults to the package palette. Default fcs_colors().

Details

WHAT TO LOOK FOR. The question this figure answers is whether the spread BETWEEN samples is larger than the bar WITHIN each of them. Where it is, the variation is something the gate is measuring. Where the bars are as wide as the scatter, the variation is the gate moving, and a group difference in that population is not interpretable however small its p-value.

Controls and QC-failed samples are excluded here rather than labelled, unlike the QC figures: their percentages are fractions of an arbitrary parent slice, so plotting them next to real samples would put two different quantities on one axis.