Abstract
Untargeted mass spectrometry-based metabolomics generates large-scale fragmentation data, typically analyzed separately in positive and negative ionization modes. The fragmentation patterns of the same molecule usually capture distinct but complementary structural information across polarities. Evaluating them simultaneously, rather than in separate molecular networks, can enhance the overall informativeness and speed the analysis. In this paper, we introduce the Neutral Molecular Network (NMN) concept, a novel strategy that unifies complementary fragmentation patterns from both polarities into a single "neutral pseudo-spectra". NMNs outperformed polarity-specific networks on several large-scale publicly available MS/MS libraries in terms of chemical reliability of spectral matches and clustering capability. The improved performances were further confirmed through the analysis of a biological untargeted dataset of Alternaria fungal extracts, where this approach facilitated the identification of previously uncharacterized toxin derivatives. NMN offers a polarity-independent framework for the structural interpretation of untargeted MS/MS data. This method can improve the efficiency of metabolite annotation pipelines and can be applicable to diverse metabolomics workflows.