Branch-Wise Regularization for Heterogeneous GNN-Based MALDI-TOF AMR Prediction with Biomarker-Consensus Edges

Abstract

This paper predicts antimicrobial resistance (AMR) from matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectra using DRIAMS, evaluated across 13 species-antibiotic datasets spanning sensitive/resistant imbalance ratios of 2.95:1 to 141.21:1 and raw spectral similarities of 0.9166–0.9946. We propose HetSAGE, a heterogeneous graph neural network (GNN) that combines full-spectrum “raw” edges with consensus “biomarker” edges derived from a 4-method feature-selection vote and applies per-edge-type dropout to regularize the two views separately. HetSAGE outperforms a plain multilayer perceptron (MLP) only when spectral similarity is high, consistent with oversmoothing effects reported in the broader GNN literature; the biomarker view converges on features close to independently established clinical markers: within ±1 Da in Staphylococcus aureus, Oxacillin, and within range in Klebsiella pneumoniae, Meropenem, confirming it is not a black box. At the same time, per-edge-type dropout protects this smaller signal from dilution by regularization tuned for the noisier raw spectrum. Together, these results indicate that heterogeneous graph structure is not a universal win for MALDI-TOF AMR prediction but a conditional one, tied to a measurable property of the data, spectral similarity, rather than to architecture alone. This caveat is largely unaddressed in current DRIAMS-based GNN work and provides a concrete signal for practitioners deciding when a graph model is worth the added complexity over a plain MLP. Author Summary Diagnosing which antibiotics will work against a bacterial infection typically takes days because it requires growing the bacteria in the lab. A faster technology, MALDI-TOF mass spectrometry, produces a chemical fingerprint of the bacteria within hours, and prior work has shown that this fingerprint alone can predict antibiotic resistance using machine learning. We build on this by connecting samples with similar fingerprints into a graph and using a graph neural network to share information across similar cases. We add a second, more targeted connection type based on a small set of chemical features that a panel of methods agrees are diagnostic, and show that these features align with markers clinicians already use. We find that this graph-based approach helps most when the raw fingerprints are very similar, but offers no advantage otherwise, an important caveat for others building similar systems, since the benefit depends on a measurable property of the data rather than being universal.