Select Regions to Highlight on a Manhattan Plot

Picks the regions (genes, windows, loci) to highlight on a Manhattan plot. Returns every region whose p-value is at or below threshold; if fewer than min_n regions pass, the selection is topped up with the next-smallest p-values so the plot is never empty (useful when no region reaches genome-wide significance).

select_highlight_regions(
  df,
  pvalue_col,
  threshold,
  label_col = "label",
  min_n = 10L
)

Arguments

df

data.frame of region-level results.

pvalue_col

Character. Name of the p-value column (e.g., "GLOW_Omni").

threshold

Numeric. P-value cutoff; typically 0.05 / nrow(df).

label_col

Character. Name of the region-label column. Default "label".

min_n

Integer. Minimum number of regions to return when fewer pass threshold. Default 10L.

Value

Character vector of region labels sorted by ascending p-value, of length max(n_pass, min_n) (capped at the number of non-NA p-values).

Details

The returned labels are taken in ascending p-value order. The selection size is min(max(n_pass, min_n), n_nonNA), where n_pass is the count at or below threshold and n_nonNA is the number of non-NA p-values; the result is empty only when there are no usable p-values. Complexity is \(O(n \log n)\) from the p-value sort.

Examples

df <- data.frame(label = paste0("g", 1:5),
                 GLOW_Omni = c(1e-9, 0.2, 0.4, 0.6, 0.8))
select_highlight_regions(df, "GLOW_Omni", threshold = 1e-6, min_n = 3L)
#> [1] "g1" "g2" "g3"