Module-clock effect sizes and p-values — tage_module

Per-module statistics for the module-clock heatmaps. With standardize = TRUE (the default) the two quantities come from different models: the p-value is the estimated marginal-mean contrast of the full model fitted on the stratum, while the effect size is the coefficient of a separate two-group model fitted on values standardised to unit variance, which makes modules comparable despite their different native scales. With standardize = FALSE the estimate, standard error and p-value all come from the same contrast.

Usage

tage_module_stats(
  data,
  module_columns,
  group_column,
  reference_group,
  compare_groups = NULL,
  covariates = NULL,
  split_by = NULL,
  variance_strata = c("all_data", "subset"),
  standardize = TRUE,
  p_adjust = TAGE_P_ADJUST_METHODS,
  p_adjust_scope = c("across_columns", "within_column", "global", "none"),
  conf_level = 0.95
)

Arguments

data: Data frame with one column per module clock.
module_columns: Character vector of module-clock columns.
group_column: Column defining the experimental groups.
reference_group: Reference level of group_column.
compare_groups: Levels compared against the reference. Default NULL uses every other level.
covariates: Covariate columns. Default NULL.
split_by: Column to stratify on. Default NULL.
variance_strata: "all_data" (default, matching the application) fits the p-value model on every group in the stratum; "subset" restricts it to the compared groups.
standardize: Whether to report standardised effect sizes. Default TRUE.
p_adjust: Method passed to p.adjust.
p_adjust_scope: Correction family. Default "across_columns" corrects across modules within each comparison and stratum, which is the application's "per group" setting; "global" matches its "globally" setting.
conf_level: Two-sided confidence level for ci_low / ci_high.

Value

A data frame with module, split, group1, group2, n, estimate, se, ci_low, ci_high, statistic, df, p_value, p_adjusted and label.

Examples

if (FALSE) { # \dontrun{
tage_module_stats(
  results,
  module_columns  = grep("^module_", names(results), value = TRUE),
  group_column    = "Genotype",
  reference_group = "WT",
  split_by        = "Tissue"
)
} # }