Back

A searchable metadata network graph for microbiome metabolomics

Charron-Lamoureux, V.; Xing, S.; Patan, A.; Walker, C.; Almada-Monter, R.; Abiead, Y.; Zhao, H. N.; Patel, L.; Weng, Y.; Gonzalez, A.; Ackermann, G.; Deleray, V.; Gandhi, V.; Mohanty, I.; Caraballo Rodriguez, A. M.; Kvitne, K. E.; Zuffa, S.; Norman, A.; Martin, A.; Chin, L.; Paz-Gonzalez, R.; Sala-Climent, M.; Suryawinata, N.; Zemlin, J.; Gouda, H.; Hu, Z.; Norton, G. J.; Rajkumar, P.; Molina, A. J. A.; Bergstrom, J.; Pinner, M.; Giddings, S.; Aron, A.; Liang, L.; Dahesh, S.; Lamichhane, S.; Reilly, E. R.; Nizet, V.; Skrip, A. E.; Lukowski, A. L.; Shore, S. F. H.; Ghoshal, S.; Engevik, M. A.;

2026-02-05 biochemistry
10.64898/2026.02.04.703849 bioRxiv
Show abstract

Establishing the biological context of microbial metabolites remains a major challenge. We present microbiomeMASST, a metadata-driven network graph that maps metabolites across 467 available datasets with 144,424 mass spectrometry files from humans, animals, and microbial culture systems. MicrobiomeMASST integrates monocultures, synthetic communities, and host-associated samples across multiple body sites and plants. MS/MS spectra can be queried to trace occurrence across hosts, experimental conditions, and interventions, enabling cross-study integration. We demonstrate this framework by contextualizing microbial-conjugated bile acids and interrogating microbiome-mediated drug metabolism. Screening gut bacteria revealed deprolylation of the angiotensin-converting enzyme (ACE) inhibitor prodrug enalapril. Using microbiomeMASST, we traced this metabolite across human cohorts, microbial isolates, environmental samples, and in Gorilla gorilla. Structural modeling and enzymatic assays showed that microbial deprolylation abolishes ACE inhibition, thereby inactivating its therapeutic effect. Together, microbiomeMASST links MS/MS spectra to biological context, converting isolated observations into an interpretable microbiome map for cross-study analysis.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.