# Cross ionization mode chemical similarity prediction between tandem mass spectra in metabolomics

## TL;DR

This paper introduces a machine learning model that improves the comparison of metabolite data across different ionization modes in mass spectrometry.

## Contribution

The novel contribution is a machine learning model, MS2DeepScore 2.0, that enables cross-ionization mode chemical similarity prediction.

## Key findings

- MS2DeepScore 2.0 successfully enables cross-ionization mode molecular networking.
- The model includes a quality estimation method to flag unreliable predictions.
- The tool enhances metabolite annotation and data exploration in untargeted metabolomics.

## Abstract

Mass spectrometry is a cornerstone of untargeted metabolomics, enabling the characterization of metabolites in both positive and negative ionization modes. However, comparisons across ionization modes have remained a substantial challenge due to the distinct fragmentation patterns produced by each polarity. To overcome this barrier, we present MS2DeepScore 2.0, a machine learning-based model to predict chemical similarity between mass fragmentation spectra, which works both between different and the same ionization modes. We demonstrate the utility of MS2DeepScore 2.0 in three case studies, where MS2DeepScore enabled cross-ionization mode molecular networking, enhancing data exploration and metabolite annotation. To ensure robustness, we have implemented a quality estimation method that flags spectra with low information content or those dissimilar to the training data, thereby minimizing false predictions. Altogether, MS2DeepScore 2.0 extends our current capabilities in organizing, exploring, and annotating untargeted metabolomics profiles.

Mass spectrometry is a cornerstone of untargeted metabolomics, but comparisons across ionization modes have remained a substantial challenge due to the distinct fragmentation patterns produced by each polarity. Here, the authors present MS2DeepScore 2.0, a machine learning-based model to predict chemical similarity between mass fragmentation spectra, which works both between different and the same ionization modes.