Compare drug sensitivities between two OncoExperiment groups — compare_drug_sensitivities,OncoExperiment-method • OncoProfilingTools

Compares PRISM drug response values between two user-defined groups stored in a single `OncoExperiment` object using per-drug Welch two-sample t-tests.

Usage

# S4 method for class 'OncoExperiment'
compare_drug_sensitivities(
  object,
  group1,
  group2,
  unit = NULL,
  p_adj_method = "BH",
  effect_threshold = 0,
  ...
)

Arguments

object: An `OncoExperiment` object with a top-level `group` column.
group1: Label for the first group of models/samples.
group2: Label for the second group of models/samples.
unit: Optional character scalar describing the expected response unit (for example, `"LFC"`). If `NULL`, the unit is inferred from the assay metadata when available and normalized to `"LFC"` when the metadata uses an equivalent label such as `"logfoldchange"`.
p_adj_method: Method passed to `stats::p.adjust()`.
effect_threshold: Non-negative numeric threshold for `abs(mean_diff)` when computing significance flags.
...: Additional arguments. Currently unused.

Value

A `DrugSensitivityComparison` object.

Details

**User is expected to define the groups prior to calling this method.** For example, to compare drug sensitivities between two cancer types, subset the `OncoExperiment` and then assign groups: `COAD$group <- c(1, 1, 2, 1, 2, 2, 1, ...)`

Whatever labels were chosen for the groups must be passed into the function. The method looks in the top-level `$group` column and compares the selected values within the PRISM assay.

This method expects a loadable `DrugResponseAssay` (typically named "PRISM") with assay data named "response".

The sign convention is: `mean_diff = mean(group1) - mean(group2)`. Negative values indicate lower average response in `group1`.