API Reference

Complete reference for all public classes and functions in MaldiBatchKit, organised by module.

Correctors

Every corrector exposes the scikit-learn fit / transform / fit_transform API, accepts batch and covariates at construction time, and aligns them to X.index at each call - so the same object is safe inside Pipeline and cross_val_score without leakage.

Base Class

maldibatchkit.BaseBatchCorrector

Base class for batch correctors that store batch at construction.

Subclass this to ship a custom corrector. See Extending MaldiBatchKit for a walkthrough.

ComBat Family

maldibatchkit.ComBat	Pipeline-friendly wrapper around ComBatModel.
maldibatchkit.SpeciesAwareComBat	ComBat-Fortin preset with species as a protected biological covariate.
maldibatchkit.QualityWeightedComBat	Weighted empirical-Bayes extension of Johnson-ComBat.

Linear / Non-Parametric

maldibatchkit.Limma

Linear-model batch subtraction following limma::removeBatchEffect.

Single-Cell-Style Integration

maldibatchkit.Harmony

Sklearn-compatible Harmony with a closed-form transform.

Simple Baselines

maldibatchkit.MedianCentering	Subtract per-batch medians from each feature.
maldibatchkit.ZScorePerBatch	Per-batch z-score normalisation.
maldibatchkit.ReferenceScaling	Rescale each batch so its per-feature mean matches a reference batch.

MALDI-Specific Corrections

maldibatchkit.BatchAwareWarping

Per-batch m/z warping to a single global reference.

Diagnostics

Generic Batch-Mixing Metrics

maldibatchkit.diagnostics.silhouette_batch	Silhouette coefficient using batch labels as clusters.
maldibatchkit.diagnostics.kbet	k-nearest-neighbours Batch Effect Test (kBET; Büttner et al. 2019).
maldibatchkit.diagnostics.lisi	Local Inverse Simpson's Index for batch mixing.

MALDI-Specific Metrics

maldibatchkit.diagnostics.peak_position_drift	Per-batch peak-position drift relative to a global reference.
maldibatchkit.diagnostics.tic_cov_per_batch	Per-batch coefficient of variation of the Total Ion Count.
maldibatchkit.diagnostics.per_batch_spectrum_count	Return the number of spectra per batch as a sorted Series.

Combined Report

maldibatchkit.diagnostics.diagnostic_report

Run every diagnostic on a (before, after) pair.

Benchmark

maldibatchkit.diagnostics.BatchCorrectionBenchmark

Diagnostic comparison of multiple batch correctors.

Metrics

Batch-aware downstream classifier metrics for model selection that generalises across sites. See Metrics Module for the full reference.

Per-batch Metric Functions

maldibatchkit.batch_roc_auc_score	Per-batch AUROC, aggregated with weights.
maldibatchkit.batch_average_precision_score	Per-batch average precision (PR-AUC), aggregated with weights.
maldibatchkit.batch_balanced_accuracy_score	Per-batch balanced accuracy, aggregated with weights.
maldibatchkit.batch_matthews_corrcoef	Per-batch Matthews correlation coefficient, aggregated with weights.
maldibatchkit.batch_f1_score	Per-batch F1 score, aggregated with weights.
maldibatchkit.batch_precision_score	Per-batch precision, aggregated with weights.
maldibatchkit.batch_recall_score	Per-batch recall, aggregated with weights.

Scorer Factory

maldibatchkit.make_batch_scorer

Return a sklearn-compatible scorer(estimator, X, y).

Visualization

maldibatchkit.viz.plot_batch_umap	Plot UMAP embeddings of X before and after batch correction.
maldibatchkit.viz.plot_peak_shift	Overlay per-batch median spectra against a reference spectrum.
maldibatchkit.viz.plot_diagnostic_summary	Plot before/after diagnostic values, one subplot per metric.

Integrations

maldibatchkit.integrations.MaldiSetAdapter

Bridge maldiamrkit.MaldiSet and MaldiBatchKit correctors.