GeneRankFit-accessors   Accessors for GeneRankFit
align_datasets          Align and merge expression + metadata (genes by
                        intersection; strict sample match)
apply_calibration       Apply stored calibration to a numeric vector of
                        probabilities
calibrate_oof           Calibrate out-of-fold probabilities (isotonic
                        or Platt)
factor_dependence       Covariate dependence of gene contributions
                        (SHAP/proxy; "expr" assay)
id_map                  Map gene identifiers using AnnotationDbi
plot_confusion_heatmap
                        Confusion-matrix heatmap
plot_embed              Decision-space embedding (OOF probability
                        space): PCA or UMAP
plot_embed_expr         Expression-space embedding (PCA or UMAP) of top
                        RF genes
plot_importance         Feature importance (top genes)
plot_roc                ROC curve from OOF probabilities (single model)
plot_roc_multi          Multi-model ROC (guardrailed OOF)
plot_shap_dependence    SHAP dependence scatter for one gene
                        (cached-first, robust, fast)
plot_sign_importance    Signed feature importance (directional effect)
prepare_data            Prepare matrices + metadata: align, (log)
                        transform, batch correct, prefilter
rank_genes              Rank genes with batch-aware cross-validation
                        (UMAP-free)
rfgr_crossval           Batch-aware cross-validation with optional
                        ComBat modes (incl. frozen ComBat)
rfgr_plot_suite         One-call plot suite (optional file export)
shap_train_ranger       Compute a SHAP matrix using a single ranger
                        model
sign_importance         Signed feature importance for RFGeneRank models
top_genes               Extract top predictive genes from a GeneRankFit
validate_genes          Validate a ranked gene set with alternate
                        learners via k-fold CV
