Test all treatments against a control group

Runs cr_test() pairwise for every treatment versus the control and adjusts the p-values across the whole family of comparisons. Both a Bonferroni and a Benjamini-Hochberg adjustment are reported next to the unadjusted p-value, never in place of it.

Usage

cr_test_all(
  experiment,
  channel,
  control_group,
  tests = "mann_whitney",
  p_adjust = "BH",
  level = c("replicate", "cell", "both")
)

Arguments

experiment

A cr_experiment.

channel

Channel name.

control_group

Control group.

tests

Test to run (see cr_test()). Only the first element is used; the argument is a vector for backwards compatibility.

p_adjust

P-value adjustment method used for the p_adj column (see stats::p.adjust). The dedicated p_bonferroni and p_BH columns are always added as well.

level

"cell", "replicate" or "both".

Value

A named list of cr_result objects, one per treatment, carrying the attributes summary (a tibble with treatment, log2_fc, p_value, cohens_d, p_adj, p_bonferroni, p_BH and interpretation), control_group, channel and level.

See also

cr_test(), cr_effect_grid().

Other statistics: cr_test()

Examples

# \donttest{
exp <- cr_example_experiment(seed = 1, n_cells_per_well = 20)
all_res <- cr_test_all(exp, channel = "marker_1",
                       control_group = "Untreated",
                       level = "replicate")
attr(all_res, "summary")
#> # A tibble: 5 × 8
#>   treatment log2_fc p_value cohens_d   p_adj p_bonferroni    p_BH interpretation
#>   <chr>       <dbl>   <dbl>    <dbl>   <dbl>        <dbl>   <dbl> <chr>         
#> 1 PosContr…   3.40  1.54e-6   2.12   4.66e-6   0.00000772 4.66e-6 strong        
#> 2 Compound…   1.07  3.12e-5   0.469  5.20e-5   0.000156   5.20e-5 weak          
#> 3 Compound…   3.05  1.86e-6   2.08   4.66e-6   0.00000932 4.66e-6 strong        
#> 4 CompoundB   0.495 2.73e-3   0.0445 2.73e-3   0.0137     2.73e-3 weak          
#> 5 CompoundC   0.923 8.20e-5   0.487  1.02e-4   0.000410   1.02e-4 weak          
# }