MaldiSuite

A Python ecosystem for MALDI-TOF spectral processing and analysis in antimicrobial resistance research

MaldiSuite combines preprocessing, batch effect correction, and deep learning classifiers in a single sklearn-compatible workflow for clinical microbiology and computational biology research.

Landing page: MaldiSuite

What this repository contains

This repository serves two purposes:

1. Landing page for the website.
2. Meta-package: a minimal PyPI package (maldisuite) that installs the three packages of the suite with pinned compatible versions.

This repository does not contain the source code of the three packages. Each package lives in its own repository, with its own issues, documentation, and release cycle.

The three packages

MaldiAMRKit owns the data model, preprocessing, evaluation, and biomarker / drift modules; MaldiBatchKit re-exports every corrector (ComBat variants, Limma, Harmony, MALDI-specific harmonisers) at the top level; MaldiDeepKit ships scikit-learn-compatible deep classifiers (MaldiMLPClassifier, MaldiCNNClassifier, MaldiResNetClassifier, MaldiTransformerClassifier).

Logo	Package	Purpose	Repository
	MaldiAMRKit	Preprocessing, evaluation, biomarkers, drift monitoring	MaldiAMRKit
	MaldiBatchKit	Batch effect correction for mass spectra	MaldiBatchKit
	MaldiDeepKit	Deep learning classifiers for MALDI-TOF spectra	MaldiDeepKit

Installation

Install the full suite with a single command:

pip install maldisuite

This installs maldiamrkit, maldibatchkit, and maldideepkit at compatible versions.

Alternatively, install individual packages separately:

pip install maldiamrkit     # preprocessing, biomarkers, drift
pip install maldibatchkit   # batch effect correction
pip install maldideepkit    # deep learning classifiers

Quick start

from maldiamrkit import MaldiSet
from maldiamrkit.evaluation import CaseGroupedKFold, vme_scorer
from maldibatchkit import BatchAwareWarping, ComBat
from maldideepkit import MaldiMLPClassifier
from sklearn.pipeline import Pipeline
from sklearn.model_selection import cross_val_score

data = MaldiSet.from_directory("driams/", "meta.csv", bin_width=3)
batch = data.meta["batch"]

pipe = Pipeline([
    ("warp",   BatchAwareWarping(batch=batch)),
    ("combat", ComBat(batch=batch, method="fortin")),
    ("clf",    MaldiMLPClassifier(input_dim=data.X.shape[1], random_state=0)),
])

scores = cross_val_score(
    pipe, data.X, data.get_y_single("Drug"),
    scoring=vme_scorer,
    cv=CaseGroupedKFold(n_splits=5),
    groups=data.meta["patient_id"],
)

Citation

Citation will be available soon.

License

MIT License - see LICENSE for the full text.

Acknowledgments

MaldiSuite builds on the work of the mass spectrometry and clinical microbiology communities. In particular, we acknowledge:

• MALDIquant (Gibb & Strimmer, 2012, Bioinformatics)
• The Borgwardt Lab and the DRIAMS benchmark (Weis et al., 2022, Nature Medicine), which established Python-based MALDI-TOF AMR prediction as a research area