# Integrated analysis of MALDI-TOF MS and whole-genome sequencing for subtyping Salmonella

## TL;DR

This study combines MALDI-TOF MS and whole-genome sequencing to rapidly subtype Salmonella, improving diagnostic efficiency and outbreak investigation.

## Contribution

A novel machine learning-based strategy integrating MALDI-TOF MS and genomic data for rapid Salmonella subtyping is proposed.

## Key findings

- Machine learning models achieved AUC values over 0.90 for key Salmonella subtypes.
- Discriminatory mass peaks correlate with genomic features, enabling rapid classification.
- The method acts as a high-throughput pre-filter to prioritize high-risk clones for WGS.

## Abstract

Current subtyping methods are often restricted by labor intensity and high costs. To address this, this study integrated matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) with whole-genome sequencing (WGS) to characterize Salmonella isolates and investigate the correlation between spectral features and genomic data. Between 2023 and 2024, 96 Salmonella isolates from Yixing, Jiangsu Province, China, underwent serotyping, WGS, and MALDI-TOF MS profiling. Serotyping and Multilocus Sequence Typing (MLST) analysis resolved 25 serovars and 21 sequence types. Machine learning models based on spectral features achieved area under the curve (AUC) values exceeding 0.90 for Salmonella typhimurium, ST11, ST155, ST19, and ST34. Specific discriminatory mass peaks were identified, and their correlations with genomic annotations were investigated through peak-gene co-occurrence analysis. The findings indicate that discriminatory MALDI-TOF MS peaks can serve as statistical indicators for specific genomic features, reflecting underlying genomic differences. This study proposes a machine learning-based classification strategy that enables rapid analysis of MALDI-TOF MS spectra in routine diagnostics, thereby extending the application of mass spectrometry in Salmonella subtyping. This strategy functions as a high-throughput pre-filter to concentrate WGS efforts on high-risk clones for accelerated outbreak investigation.