MaldiAMRKit

A comprehensive toolkit for MALDI-TOF mass spectrometry data preprocessing for antimicrobial resistance (AMR) prediction purposes

Installation • Features • Quick Start • Documentation • Tutorials • Contributing • Citing • License

Installation

pip install maldiamrkit

Optional: Batch Correction & UMAP

pip install maldiamrkit[batch]

Installs combatlearn for ComBat-based batch effect correction and umap-learn for UMAP exploratory plots.

Development Installation

git clone https://github.com/EttoreRocchi/MaldiAMRKit.git
cd MaldiAMRKit
pip install -e .[dev]

Features

Preprocessing

• Composable Pipeline: Build custom PreprocessingPipeline from individual transformers (smoothing, baseline correction, normalization, trimming), serializable to JSON/YAML
• Multiple Binning Strategies: Uniform, proportional, adaptive, and custom bin edges
• Quality Metrics: SNR estimation, comprehensive quality reports, and alignment assessment
• Replicate Merging: Mean/median/weighted merging with correlation-based outlier detection

Alignment & Detection

• Spectral Alignment: Shift, linear, piecewise, and DTW warping for both binned and raw full-resolution spectra
• Peak Detection: Local maxima and persistent homology methods

Evaluation

• AMR Metrics: VME, ME, sensitivity, specificity, categorical agreement, and amr_classification_report following EUCAST/CLSI conventions
• Label Encoding: LabelEncoder for mapping R/I/S to binary with configurable intermediate handling
• Stratified Splitting: Species-drug stratified and case-based (patient-grouped) splitting to prevent data leakage

Differential Analysis

• DifferentialAnalysis: Per-bin statistical testing (Mann-Whitney U, Welch's t-test) between resistant and susceptible groups, with multiple-testing correction, log2 fold change, and Cohen's d effect size
• Peak Selection: top_peaks() by adjusted p-value, significant_peaks() with fold-change and p-value thresholds, compare_drugs() for multi-drug boolean significance matrices
• AMR-Aware Plots: plot_volcano(), plot_manhattan() along the m/z axis, and plot_drug_comparison() with binary heatmap or UpSet-style intersection view

Drift Monitoring

• DriftMonitor: Anchor a baseline on early timestamps (default: first 20%) and track temporal drift via three complementary views - reference similarity of per-window median spectra, PCA centroid trajectory in a baseline-fitted PCA space, and Jaccard stability of top-k differential peaks over time
• Trajectory Plots: plot_reference_drift, plot_pca_drift, plot_peak_stability, plot_effect_size_drift

Data Management

• Dataset Building & Loading: DatasetBuilder and DatasetLoader with pluggable layout adapters (FlatLayout, BrukerTreeLayout, DRIAMSLayout, MARISMaLayout)
• Bruker Format Support: Read Bruker flexAnalysis binary data (fid/1r + acqus) natively via read_spectrum() on directories
• MIC Parsing: parse_mic_column() for parsing MIC strings with qualifiers and European decimals
• Composable Filters: SpeciesFilter, DrugFilter, QualityFilter, MetadataFilter combinable with &/|/~ operators
• Spectrum Export: Save spectra to CSV or TXT via MaldiSpectrum.save() and MaldiSet.save_spectra()

Visualization & Tools

• Exploratory Plots: PCA, t-SNE, and UMAP scatter plots colored by species, resistance phenotype, or any metadata column
• Batch Effect Correction: Multi-site/multi-instrument correction via combatlearn (pip install maldiamrkit[batch])
• CLI: maldiamrkit preprocess, maldiamrkit quality, and maldiamrkit build for batch processing
• Parallel Processing: Multi-core support via n_jobs parameter
• ML-Ready: Direct integration with scikit-learn pipelines

Documentation

Full documentation is available at maldiamrkit.readthedocs.io.

Quick Start

Load and Preprocess a Single Spectrum

from maldiamrkit import MaldiSpectrum

# Load spectrum from file
spec = MaldiSpectrum("data/spectrum.txt")

# Preprocess: smoothing, baseline removal, normalization
spec.preprocess()

# Optional: bin to reduce dimensions
spec.bin(bin_width=3)  # 3 Da bins

# Visualize
from maldiamrkit.visualization import plot_spectrum
plot_spectrum(spec, binned=True)

Build a Dataset from Multiple Spectra

from maldiamrkit import MaldiSet

# Load multiple spectra with metadata
data = MaldiSet.from_directory(
    spectra_dir="data/spectra/",
    meta_file="data/metadata.csv",
    aggregate_by=dict(antibiotics="Drug", species="Escherichia coli"),
    bin_width=3
)

# Access features and labels
X = data.X  # Feature matrix
y = data.get_y_single("Drug")  # Target labels

Machine Learning Pipeline

from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import cross_val_score
from maldiamrkit.alignment import Warping
from maldiamrkit.detection import MaldiPeakDetector

# Create ML pipeline
pipe = Pipeline([
    ("peaks", MaldiPeakDetector(binary=False, prominence=0.05)),
    ("warp", Warping(method="shift")),
    ("scaler", StandardScaler()),
    ("clf", RandomForestClassifier(n_estimators=100, random_state=42))
])

# Cross-validation
scores = cross_val_score(pipe, X, y, cv=5, scoring="accuracy")
print(f"CV Accuracy: {scores.mean():.3f} +/- {scores.std():.3f}")

For more examples covering alignment, filtering, evaluation, CLI usage, and more, see the Quickstart Guide and API Reference.

Tutorials

For more detailed examples, see the notebooks:

• Quick Start - Loading, preprocessing, binning, and quality assessment
• Peak Detection - Local maxima and persistent homology methods
• Alignment - Warping methods and alignment quality
• Evaluation - AMR metrics, label encoding, and stratified splitting
• Exploration - PCA, t-SNE, UMAP visualizations and batch correction
• Differential Analysis - R vs. S peak testing, volcano/Manhattan plots, and multi-drug comparison
• Drift Monitoring - Baseline-anchored drift detection: reference similarity, PCA trajectory, peak stability, and effect-size drift

Contributing

Pull requests, bug reports, and feature ideas are welcome. See the Contributing Guide for how to get started.

Citing

If you use MaldiAMRKit in your research, please cite:

Rocchi, E., Nicitra, E., Calvo, M. et al. Combining mass spectrometry and machine learning models for predicting Klebsiella pneumoniae antimicrobial resistance: a multicenter experience from clinical isolates in Italy. BMC Microbiol (2026). doi:10.1186/s12866-025-04657-2

See the full publications list for more papers using MaldiAMRKit.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgements

This toolkit is inspired by:

Weis, C., Cuénod, A., Rieck, B., et al. (2022). Direct antimicrobial resistance prediction from clinical MALDI-TOF mass spectra using machine learning. Nature Medicine, 28, 164-174. https://doi.org/10.1038/s41591-021-01619-9