Skip to contents

Plot Enrichment Results

Source: R/11_viz_statistical.R
av_plot_enrichment.Rd

Unified entry point for Fisher's-Exact-Test enrichment visualizations. style = "bar" ranks groups by significance, "volcano" plots effect size (Log2 OR) against significance, and "threshold_scan" sweeps the hit/non-hit quantile cutoff and plots the log2(OR) trajectory per group. The "bar" and "volcano" styles read the stored enrichment and require av_calc_enrichment to have been run; "threshold_scan" runs its own enrichment sweep with return_only = TRUE and does not mutate obj$stats.

Usage

av_plot_enrichment(
  obj,
  style = c("bar", "volcano", "threshold_scan"),
  group_by = "output_type",
  fdr_threshold = 0.05,
  top_n = 20L,
  label_top_n = 15L,
  highlight = NULL,
  highlight_color = "#FF6600",
  thresholds = seq(0.5, 0.95, 0.05),
  scan_top_n = NULL,
  key = "enrichment"
)

Arguments

obj

An AlphaVarSet object.

style

One of "bar", "volcano", "threshold_scan".

group_by

Grouping column for the "threshold_scan" sweep (default "output_type").

fdr_threshold

Adjusted-P cutoff for "bar"/"volcano" significance (default 0.05).

top_n

Maximum number of groups for the "bar" style (default 20).

label_top_n

Number of points to label in the "volcano" (default 15).

highlight

Character vector of group names to ring and label in the "volcano" (optional).

highlight_color

Colour of the volcano highlight ring (default "#FF6600").

thresholds

Numeric vector of quantile thresholds for the "threshold_scan" sweep (default seq(0.5, 0.95, 0.05)).

scan_top_n

Integer or NULL. For "threshold_scan", keep only the scan_top_n groups with the largest max(abs(log2OR)).

key

Slot in obj$stats$comparisons the "bar" and "volcano" styles read (default "enrichment"). Must match the key passed to av_calc_enrichment. Ignored by "threshold_scan", which computes its own sweep.

Value

A ggplot2 object.

Interpretation

Bars/points beyond the significance line are enriched or depleted (red = more high-scoring variants than expected, blue = fewer). In the volcano, large points have more statistical power; labelled points are the top hits. In the threshold scan, flat trajectories indicate threshold-robust enrichment, while steeply rising or falling curves mean the signal depends on where the cutoff is placed.

See also

av_calc_enrichment

Examples

example_file <- system.file("extdata", "bcl11a_rs1427407_example.csv", package = "AlphaVaR")
obj <- av_import_csv(example_file)
#> Reading bcl11a_rs1427407_example.csv ...
obj <- av_calc_enrichment(obj)
av_plot_enrichment(obj, style = "bar")

av_plot_enrichment(obj, style = "volcano", label_top_n = 5)
#> Fewer than 15 enrichment groups. Consider style = "bar" for small group counts.


# A second enrichment lives under its own key and is plotted through it.
obj <- av_calc_enrichment(obj, group_by = "biosample_type",
                          key = "enrichment_by_biosample_type")
#> Warning: Dropped 4 of 2676 rows with NA in 'biosample_type' or 'quantile_score'.
av_plot_enrichment(obj, style = "volcano",
                   key = "enrichment_by_biosample_type")
#> Fewer than 15 enrichment groups. Consider style = "bar" for small group counts.