Decoding Microbial Messages: Untargeted Metabolomics and Data Science-Driven Discovery

Ipsita Mohanty
Department of Veterinary and Biomedical Sciences, The Pennsylvania State University, University Park, PA, USA

Ipsita Mohanty, PhD (Presenter) Penn State University

Relevant Financial Disclosures (within past 24 months, reported on Apr 22, 2026) No relevant financial relationship(s) to disclose.

What if the gut microbiome communicates with the host through a hidden chemical language - encoded not only in the genes, but also in the metabolites? Bile acids are emerging as compelling candidates for this role. Traditionally viewed as digestive detergents, bile acids are now recognized as potent signaling molecules regulating metabolism, immunity, and systemic physiology. Yet, this view captures only a fraction of their true complexity. Microbial metabolism transforms host-derived bile acids into a previously underappreciated chemical space. In this keynote, I will present a data science-driven framework to decode this hidden chemical language. By combining untargeted LC-MS/MS with reusable, bile acid-selective MS/MS filters, we enable large-scale mining of public metabolomics repositories to systematically discover and resolve bile acid isomers. This approach reveals thousands of previously uncharacterized bile acids, including novel conjugates shaped by diet and microbial metabolism, expanding the known bile acid universe far beyond canonical pathways.

To translate discovery into biological insight, we integrate this data with artificial intelligence. Large language models (LLMs) are used to synthesize information across public datasets, metadata, and literature, constructing a bile acid-centered knowledge atlas that links molecular structures to tissues, diseases, dietary exposures, and microbial origins. Together, these advances support a new conceptual model: bile acids as molecular carriers of microbial messages, where isomerization and conjugation encode functional information transmitted to the host. By decoding this language, we open new opportunities for biomarker discovery, mechanistic insight, and therapeutic intervention.
This work highlights how the convergence of high-resolution mass spectrometry, repository-scale data mining, and AI is redefining our ability to interpret the metabolome and reveals that some of the most important signals in human health may be hidden in plain sight.