Clinical variables against population frequencies and marker intensities

MARKER INTENSITY IS TESTED WITHIN A POPULATION. A marker's level is a property of a defined cell population, which is how a differential-state test asks it (diffcyt-DS, Weber et al. 2019; stats_marker_state() here), so each population and marker pair is one test, keyed "<population>, <marker>". This used to take, per sample and marker, the median of the declared populations' medians: populations nest and the list is the specification's, so that value followed whichever population type the specification listed most (for HLA-DR, the T-cell subsets) and described no set of cells. The pipeline passes functional_markers.csv, whose pairs are the ones the gating document scopes as interpretable, and population_marker_mfi.csv only where no functional block resolved.

Usage

stats_clinical_association(
  freq,
  mfi = NULL,
  clin = list(),
  patient_of = NULL,
  value_col = NULL,
  min_n = 4L
)

Arguments

freq

population_frequencies table.

mfi

per-sample marker medians by population, with sample_id, population, marker and median_asinh (functional_markers.csv or population_marker_mfi.csv), optional.

clin

named list of clinical variables, sample_id -> value.

patient_of

named vector sample_id -> patient_id, used to report an association at the donor level where a variable is a property of the donor rather than of the acquisition. NULL treats every row as its own subject.

value_col

the frequency column to use.

min_n

fewest samples for a test.

Value

list(populations, markers)