OmicStackedEnsemble R6 Class

Orchestrates multi-omics stacking with leakage-free OOF prediction generation. Trains base learners per modality and combines using a meta-learner.

Based on Super Learner methodology (van der Laan et al., 2007).

References

van der Laan, M.J., Polley, E.C., Hubbard, A.E. (2007). Super Learner. Statistical Applications in Genetics and Molecular Biology, 6(1).

Public fields

modalities

List of OmicModalitySpec objects

meta_learner

Learner for combining base predictions

resampling

Resampling strategy for OOF predictions

prediction_type

"prob" for probabilities, "response" for class labels

task_type

"classif" or "regr"

Methods

Public methods

• OmicStackedEnsemble$new()

• OmicStackedEnsemble$train()

• OmicStackedEnsemble$predict()

• OmicStackedEnsemble$is_fitted()

• OmicStackedEnsemble$get_oof_stack()

• OmicStackedEnsemble$get_modality_weights()

• OmicStackedEnsemble$print()

• OmicStackedEnsemble$clone()

---

Method new()

Create a new stacked ensemble

Usage

OmicStackedEnsemble$new(
  modalities,
  meta_learner = NULL,
  resampling = NULL,
  prediction_type = "prob",
  task_type = "classif"
)

Arguments

modalities

List of OmicModalitySpec objects

meta_learner

Learner for stacking (default: logistic regression)

resampling

Resampling strategy (default: 5-fold CV)

prediction_type

"prob" or "response"

task_type

"classif" or "regr"

Returns

An OmicStackedEnsemble object

---

Method train()

Train the stacked ensemble

Usage

OmicStackedEnsemble$train(y, target_name = "target")

Arguments

y

Target vector (named with sample IDs)

target_name

Name for target column

Returns

Self (invisibly), for method chaining

---

Method predict()

Predict using the stacked ensemble

Usage

OmicStackedEnsemble$predict(newdata_list)

Arguments

newdata_list

Named list of data matrices, one per modality. Names must match modality IDs. Missing modalities can be NULL.

Returns

Prediction object

---

Method is_fitted()

Check if ensemble is fitted

Usage

OmicStackedEnsemble$is_fitted()

Returns

Logical

---

Method get_oof_stack()

Get OOF stacking features (for auditing)

Usage

OmicStackedEnsemble$get_oof_stack()

Returns

Data frame with OOF predictions used for meta-learner training

---

Method get_modality_weights()

Get modality contribution (meta-learner coefficients if linear)

Usage

OmicStackedEnsemble$get_modality_weights()

Returns

Named vector or NULL if meta-learner is not linear

---

Method print()

Print method

Usage

OmicStackedEnsemble$print()

---

Method clone()

The objects of this class are cloneable with this method.

Usage

OmicStackedEnsemble$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.