Single-sample matched-null benchmark for panel-vs-random-panel AUC inference

Generalised matched-null benchmark introduced in the single-sample scoring bank (Module A validation framework). Unlike the simpler os_panel_null_benchmark in panel-gates.R, this implementation stratifies random-panel draws by per-feature detection-rate and log-mean-abundance quartile bins, explicitly excludes hemolysis-marker miRNAs when post_hemolysis_corrected = TRUE, and provides a three-tier fallback protocol for small platforms with exhausted candidate pools. BH-FDR and block-aware BH-FDR correction helpers are also provided for multi-modality family testing.

Methods provided:

• singlesample_matched_null_benchmark: main benchmark function.

• singlesample_hanley_mcneil_auc_ci: closed-form Hanley-McNeil 1982 95% AUC confidence interval helper.

• singlesample_bh_fdr_correct_matched_null: BH-FDR within a modality family.

• singlesample_holm_correct_familywise: Holm correction for family-wise method × claim-state contrasts.

• singlesample_bh_fdr_correct_blocked: block-aware two-stage BH-FDR for specimen-shared cohort clusters.

References

Benjamini Y, Hochberg Y. (1995) Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society Series B 57(1): 289–300.