TRexBiobankScreeningSelector: Biobank Screening with the T-Rex Selector

TRexBiobankScreeningSelectorR Documentation

Biobank Screening with the T-Rex Selector

Description

Per phenotype the fast screening (T-Rex+Screening), and the T-Rex selector as fallback when no FDR estimate of the screening lies in the accepted range.

Decision rule per phenotype

The screening gives two FDR estimates, the one of its ordinary variant and the one of its bootstrap variant. A variant is accepted when its estimate lies in [lower_bound_FDR, upper_bound_FDR] of trex_biobank_control(). Of two accepted variants the one with the larger estimate is taken, and the bootstrap variant on a tie. With no accepted variant the T-Rex selector runs at target_FDR_trex.

Both estimates come from one set of K experiments. The screening runs once, and the bootstrap variant is derived from its coefficients.

Methods

Public methods


TRexBiobankScreeningSelector$new()

Create a new TRexBiobankScreeningSelector object.

Usage
TRexBiobankScreeningSelector$new(
  X,
  Y,
  tFDR = NULL,
  seed = -1,
  verbose = FALSE,
  biobank_control = trex_biobank_control(),
  screen_control = trex_screen_control(),
  control = trex_control()
)
Arguments
X

Feature matrix (n x p) in memory, or an mmap_matrix, the handle of a memory-mapped file. Both are bound without a copy. The inner selectors standardize a memory-mapped design in place per call and restore it, which needs "readwrite" mode. Under scaling_mode = "NONE" they write nothing and "readonly" mode suffices.

Y

Response vector (1D) or matrix (2D). The matrix is in memory or an mmap_matrix, the handle of a memory-mapped file, and is bound without a copy. It is only read, so "readonly" mode suffices for it.

tFDR

Target FDR level of the fallback T-Rex selector. NULL (default) takes target_FDR_trex of biobank_control. A supplied value replaces it, as in TRexSelector 2.0.1.

seed

Random seed. A negative value (default: -1) gives a run that is not reproducible. Phenotype k (the k-th column of Y) runs with seed + k - 1, so the phenotypes draw independent dummies and a single phenotype is reproduced by a single-phenotype run with that seed.

verbose

Whether to print progress (default: FALSE).

biobank_control

A biobank control list from trex_biobank_control() (default: trex_biobank_control()).

screen_control

A screen control list from trex_screen_control() (default: trex_screen_control()). Its use_bootstrap_CI has no effect here: the biobank screening computes the ordinary and the bootstrap variant for every phenotype.

control

A control list from trex_control() (default: trex_control()). The screening takes the values "STANDARD" (default), "SEEDED", "PERMUTATION" and "PERMUTATION_SEEDED" of lloop_strategy and refuses "HCONCAT" and "SKIPL", as TRexScreeningSelector does.


TRexBiobankScreeningSelector$select()

Run the Biobank Screening algorithm.

Usage
TRexBiobankScreeningSelector$select()
Returns

For a vector Y, a single per-phenotype record: a named list with phenotype_index, selected_indices, estimated_FDR, method_used ("Screen-TRex (ordinary)", "Screen-TRex (bootstrap-CI)" or "T-Rex (fallback)"), estimated_FDR_screen_ordinary, estimated_FDR_screen_bootstrap, selected_indices_screen_ordinary, selected_indices_screen_bootstrap, and used_fallback_trex (indices are 1-based). estimated_FDR is the estimate of the accepted screening variant, or, for a fallback result, the target level target_FDR_trex, at which the T-Rex selector controls the FDR. An estimate can equal the target, so used_fallback_trex tells the two cases apart. The two screening estimates and the two screening index vectors are stored for every phenotype, a fallback result included. For a matrix Y, a plain list of such records, one per column, so res[[i]] has the same shape as the vector case. This mirrors the Python binding, which returns one result object (vector) or a list of them (matrix). A tidy summary table is one line away, e.g. do.call(rbind, lapply(res, function(r) as.data.frame(r[c("phenotype_index", "estimated_FDR", "method_used", "used_fallback_trex")]))).


TRexBiobankScreeningSelector$clone()

The objects of this class are cloneable with this method.

Usage
TRexBiobankScreeningSelector$clone(deep = FALSE)
Arguments
deep

Whether to make a deep clone.

References

J. Machkour, M. Muma, and D. P. Palomar, "False Discovery Rate Control for Fast Screening of Large-Scale Genomics Biobanks," in Proc. 22nd IEEE Statist. Signal Process. Workshop (SSP), 2023, pp. 666-670.

F. Scheidt, J. Machkour, and M. Muma, "Solving FDR-Controlled Sparse Regression Problems with Five Million Variables on a Laptop," in Proc. IEEE 9th Int. Workshop Comput. Adv. Multi-Sensor Adapt. Process. (CAMSAP), 2023, pp. 116-120.

J. Machkour, M. Muma, and D. P. Palomar, "The Terminating-Random Experiments Selector: Fast High-Dimensional Variable Selection with False Discovery Rate Control," Signal Process., vol. 231, Art. no. 109894, 2025, doi: 10.1016/j.sigpro.2025.109894.

See Also

Other T-Rex selectors and controls: Gauss_data, TRexDASelector, TRexGVSSelector, TRexSPCASelector, TRexScreeningSelector, TRexSelector, TRexTikhonovSelector, trex_biobank_control(), trex_build_group_centering_k(), trex_control(), trex_da_control(), trex_default_lambda2_grid(), trex_dummy_distribution(), trex_fdp(), trex_fdp_dense(), trex_gamma_to_k(), trex_get_num_threads(), trex_gvs_control(), trex_gvs_lambda2_power_rule(), trex_normalize_design(), trex_precision(), trex_recall(), trex_screen_control(), trex_spca_control(), trex_tikhonov_control(), trex_tikhonov_lambda2_power_rule(), trex_tpp(), trex_tpp_dense()

Examples


data(Gauss_data)
X <- Gauss_data$X
# A vector is one phenotype and gives one record, a matrix gives a list of records
y <- drop(Gauss_data$y)
ctrl <- trex_biobank_control(target_FDR_trex = 0.1)
sel <- TRexBiobankScreeningSelector$new(X, y, biobank_control = ctrl)
res <- sel$select()
res$selected_indices



TRexSelector documentation built on Nov. 20, 2026, 5:09 p.m.