Annotating metabolite mass spectra with domain-inspired chemical formula transformers
Name
Manuscript and SI.pdf
Description
Accepted version
Size
3.07 MB
Format
Adobe PDF
Checksum (MD5)
3b63d90194050adfff7a5c83b7b728e3
Author(s)
Goldman, Samuel
Wohlwend, Jeremy
Stražar, Martin
Haroush, Guy
Xavier, Ramnik J
Coley, Connor W
Date Issued
August 17, 2023
Journal
Nature Machine Intelligence
Publisher
Springer Science and Business Media LLC
Citation
Goldman, S., Wohlwend, J., Stražar, M. et al. Annotating metabolite mass spectra with domain-inspired chemical formula transformers. Nat Mach Intell 5, 965–979 (2023).
Version
Author's final manuscript
Abstract
Metabolomics studies have identified small molecules that mediate cell signaling, competition and disease pathology, in part due to large-scale community efforts to measure tandem mass spectra for thousands of metabolite standards. Nevertheless, the majority of spectra observed in clinical samples cannot be unambiguously matched to known structures. Deep learning approaches to small-molecule structure elucidation have surprisingly failed to rival classical statistical methods, which we hypothesize is due to the lack of in-domain knowledge incorporated into current neural network architectures. Here we introduce a neural network-driven workflow for untargeted metabolomics, Metabolite Inference with Spectrum Transformers (MIST), to annotate tandem mass spectra peaks with chemical structures. Unlike existing approaches, MIST incorporates domain insights into its architecture by encoding peaks with their chemical formula representations, implicitly featurizing pairwise neutral losses and training the network to additionally predict substructure fragments. MIST performs favorably compared with both standard neural architectures and the state-of-the-art kernel method on the task of fingerprint prediction for over 70% of metabolite standards and retrieves 66% of metabolites with equal or improved accuracy, with 29% strictly better. We further demonstrate the utility of MIST by suggesting potential dipeptide and alkaloid structures for differentially abundant spectra found in an inflammatory bowel disease patient cohort.
MIT Department
Massachusetts Institute of Technology. Computational and Systems Biology Program
Massachusetts Institute of Technology. Department of Chemical Engineering
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
Abdul Latif Jameel Clinic for Machine Learning in Health
Broad Institute of MIT and Harvard
Terms of Use
Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use.
Persistent DSpace Link
DOI of Published Version
10.1038/s42256-023-00708-3
