Skip to contents

Benjamini-Yekutieli (2001) false discovery rate correction

Source: R/fdr.R
fdr_by.Rd

Thin wrapper around stats::p.adjust() with method = "BY" – base R already implements this correctly, so it is not reimplemented.

Usage

fdr_by(p, q = 0.05)

Arguments

p

Numeric vector of raw p-values in [0, 1] (may contain NA).

q

Numeric. Target FDR level – see fdr_bh()'s own q documentation for the distinction between this and a per-test alpha, which applies identically here.

Value

A list with q_value (BY-adjusted p-values) and reject (logical).

Details

Function type: Support function – computes the BY safeguard used internally by fdr_correction(). Not exported; call fdr_correction(p, method = "BY") for a BY-only result.

Typical use

Supply one family of raw p-values to fdr_correction() with method = "BY" when arbitrary dependence is a material concern, or compare it with BH as a sensitivity analysis.

Methodological details

Methods and method selection

fdr_bh() controls FDR under independence and specified forms of positive dependence. A positive Moran statistic can be compatible with that setting, but it does not prove the formal PRDS condition. BY remains valid under arbitrary dependence, at the cost of being more conservative (usually fewer rejected hypotheses) than BH for the same data. It is therefore available as an explicit safeguard when arbitrary dependence is a scientifically material concern, not as the package default. Agreement between BH and BY is a useful sensitivity result, but it does not by itself establish BH's dependence assumptions.

Statistical assumptions and limitations

BY controls FDR without requiring the independence or PRDS conditions used by BH. Its harmonic correction can be substantially conservative, so this broader guarantee may considerably reduce power.

Quality assurance

Adjusted values are generated directly by stats::p.adjust() and are also checked through the package's automated FDR tests.

References

Primary method reference:

  • Benjamini, Y., & Yekutieli, D. (2001). The control of the false discovery rate in multiple testing under dependency. Annals of Statistics, 29(4), 1165-1188. doi:10.1214/aos/1013699998

See also

Other FDR correction functions: fdr_bh(), fdr_bky(), fdr_comparison_barplot(), fdr_correction(), fdr_direction_plot(), fdr_direction_summary(), fdr_pvalue_histogram(), fdr_significance_maps(), fdr_summary(), fdr_threshold_plot()

Examples

# The same five p-values used elsewhere in this file. Called
# internally by fdr_correction() -- the public entry point is:
# fdr_correction(c(0.001, 0.01, 0.02, 0.5, 0.8), method = "BY")