Single-sample additional within-sample methods (Module A, P2)

Two additional Module-A within-sample methods introduced in the single-sample scoring bank, complementing the five core methods in singlesample-within-sample.R.

Methods provided:

• fit_logistic_normal_eb / apply_logistic_normal_eb: frozen-reference empirical-Bayes denoiser using logistic-normal posterior shrinkage (Aitchison & Shen 1980; Efron 2010). Per-feature depth-weighted CLR shrinkage toward a training-cohort prior.

• fit_frozen_quantile / apply_frozen_quantile: monotone quantile calibrator for cross-platform mapping. Frozen empirical-CDF mapping from test-sample feature values to training-distribution quantile values (Bolstad et al. 2003; Hicks et al. 2018).

References

Aitchison J, Shen SM. (1980) Logistic-normal distributions: some properties and uses. Biometrika 67(2): 261–272.

Efron B. (2010) Large-Scale Inference. Cambridge University Press.

Bolstad BM, Irizarry RA, Astrand M, Speed TP. (2003) A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19(2): 185–193.

Hicks SC, Okrah K, Paulson JN, Quackenbush J, Irizarry RA, Bravo HC. (2018) Smooth quantile normalization. Biostatistics 19(2): 185–198.