Score topological persistence-image (tda-ph) within-sample discriminator

Scores each row of X independently with the frozen model learned by fit_tda_ph. The model-universe features present in X are taken in the frozen canonical order, a per-specimen rCLR landscape is built over that row's own positive support, finite 0-dimensional lower-star classes are rasterised through the frozen exact-integral persistence-image grid, and the resulting vector is passed through the frozen standardisation and ridge-LDA head. Larger values are more case-like.

Scoring uses only each row's own values plus the frozen model – no test-batch renormalisation, no cross-row coupling, no statistic estimated from X. If fewer than model$hp$min_features model-universe features are present, the documented neutral score 0 is returned for every row; a specimen whose own positive-support count is below min_features likewise scores the neutral 0. A specimen with sufficient support but an empty finite diagram (e.g. a strictly monotone landscape) is not abstained: its all-zero persistence image is standardised and scored by the frozen head, yielding the constant empty-diagram baseline described in fit_tda_ph (which is generally not the bare intercept $b$).

Usage

score_tda_ph(model, X, meta = NULL)

Arguments

model: A tda_ph_model object returned by fit_tda_ph.
X: Numeric matrix (samples $\times$ features) of non-negative abundance values. Columns must be named feature ids.
meta: Optional per-sample metadata. Accepted for interface uniformity and ignored by this method.

Value

Plain finite numeric vector of length nrow(X). Larger values are more case-like. The score of a row depends only on that row's values and the frozen model, and is exactly invariant to per-sample positive scaling.

References

Edelsbrunner H, Letscher D, Zomorodian A. (2002) Topological persistence and simplification. Discrete & Computational Geometry 28(4): 511-533.

Adams H, Emerson T, Kirby M, et al. (2017) Persistence images: a stable vector representation of persistent homology. Journal of Machine Learning Research 18(8): 1-35.

Examples

if (FALSE) { # \dontrun{
set.seed(1)
X <- matrix(stats::rgamma(80 * 40, shape = 10, rate = 1), nrow = 80,
            dimnames = list(NULL, paste0("miR-", seq_len(40))))
y <- rep(c(0, 1), each = 40)
X[y == 1, 15:24] <- X[y == 1, 15:24] * rep(c(1.8, 0.55), 5)
model <- fit_tda_ph(X, y)
score_tda_ph(model, X[1, , drop = FALSE])
} # }