MolNetInvert: A Tool for Creating Metadata Based Molecular Networks and Their Applications in Large Microbial Datasets

Untargeted mass spectrometry-based metabolomics, crucial in diverse research areas such as clinical diagnostics and chemical ecology, often encounters the challenge of interpreting complex datasets with thousands of unique, low-annotation-rate features. Advancements in MS2-based molecular networking, particularly Feature-Based Molecular Networking and its offshoots, have improved data deconvolution and annotation. Although this is the case, these approaches often result in networks with numerous singletons, complicating interpretation when extended metadata is incorporated. Herein, we introduce Metadata-Based Molecular Networks (MBMN), a new approach to visualizing complex untargeted metabolomics datasets that enhances the interpretability of molecular networks in untargeted metabolomics, as well as MolNetInvert, an open-source GUI tool for producing these new networks. MolNetInvert simplifies network visualization by collapsing clusters into single nodes and reorganizes the networks into metadata centric networks. Overall, MBMN produces new networks based on selected metadata, creating new edges between metadata-linked nodes. Its utility is demonstrated through three case studies: 1) a co-culture analysis of metabolite production in fungal strains, 2) an investigation of secondary metabolism in black Aspergilli species using a public dataset, and 3) a comparative study of ant fungal gardens and trash piles. These applications underscore the effectiveness of MBMN’s in clarifying data analysis in mass spectrometry-based metabolomics networking, showcasing its potential in various research fields dealing with complex datasets.