Skip to content
DraconiticResearch
Search Sources API
190,869 studies31 sourcesUpdated 7 min ago
Peer-reviewedCC BYNature Communications

End-to-end multimodal structure elucidation from raw spectra combining contrastive learning and evolutionary algorithms

  • Adrian Mirza1,2
  • Luc Patiny3
  • Kevin Maik Jablonka4,5,6,7
7 affiliations
  1. 1Helmholtz Institute for Polymers in Energy Applications Jena (HIPOLE Jena), Jena, Germany. [email protected]
  2. 2Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Berlin, Germany. [email protected]
  3. 3Zakodium Sárl, Lonay, Switzerland
  4. 4Center for Energy and Environmental Chemistry Jena (CEEC Jena), Friedrich Schiller University Jena, Jena, Germany. [email protected]
  5. 5Helmholtz Institute for Polymers in Energy Applications Jena (HIPOLE Jena), Jena, Germany. [email protected]
  6. 6Jena Center for Soft Matter (JCSM), Friedrich Schiller University Jena, Jena, Germany. [email protected]
  7. 7Laboratory of Organic and Macromolecular Chemistry (IOMC), Friedrich Schiller University Jena, Jena, Germany. [email protected]
Open at sourcePDFDOICite

Abstract

Elucidating molecular structures from spectroscopic data remains one of chemistry's most fundamental challenges, typically requiring extensive expert knowledge and manual interpretation of multiple analytical techniques. This is because the structure elucidation problem often has degenerate solutions for a limited set of experimental data. Existing computational approaches are limited to single spectroscopic modalities, require extensive manual preprocessing, and lack the confidence estimates and context necessary for practical application. Here we present SECS, a framework that combines contrastive learning with evolutionary algorithms to automate structure elucidation directly from raw, multimodal spectroscopic data. By aligning embeddings across NMR, infrared, and mass spectrometry, SECS mimics how experts use multiple spectroscopic lenses while providing calibrated confidence scores and relevant database context. On challenging molecular identification tasks, SECS matches expert chemist performance in head-to-head comparisons in a pilot study. The system successfully identifies incorrect structure assignments in published literature and adapts to new chemical domains without retraining by updating its reference database. Our approach demonstrates how synergistic combination of machine learning paradigms can solve analytical bottlenecks that have constrained chemical discovery.

Related

  1. Chemistry-Informed Multimodal Model for Structure Elucidation in Automated Reaction Discovery
    Maik G. Niedziella, Philipp M. Pflüger, Ajnabiul Hoque +1
    ChemRxivPreprintCC BYFull textPhysical chemistryChemRxiv
  2. Pushing the Limits of One-Dimensional NMR Spectroscopy for Automated Structure Elucidation Using Artificial Intelligence
    Frank Hu, J. A. Tubb, Dimitris Argyropoulos +5
    Journal of Chemical Information and ModelingPeer-reviewedLicence not statedFull text7 citedPhysical chemistryOpenAlex
  3. A Modular Rapid-Injection NMR for Structure Elucidation and Kinetic Analysis.
    Andrew J Ressler, Han-Hsiang Hsu, Andy A Thomas
    Organic lettersPeer-reviewedCC BYFull textPhysical chemistryEurope PMC
  4. MAST: Motif-Augmented Diffusion with Search Tree for Spectroscopic Molecular Structure Elucidation
    Chenghao Jia, Mengdi Liu, Hong Chang +2
    arXivPreprintarXiv licenceFull textPhysical chemistryarXiv
  5. Context-Aware Automated NMR Structure Elucidation
    Frank Hu, Jonathan M. Tubb, Grant M. Rotskoff +2
    arXivPreprintCC BY 4.0Full textPhysical chemistryarXiv

Cite

BibTeXRIS
BibTeX
@article{mirza2026endtoend,
  title = {{End-to-end multimodal structure elucidation from raw spectra combining contrastive learning and evolutionary algorithms}},
  author = {Adrian Mirza and Luc Patiny and Kevin Maik Jablonka},
  journal = {Nature Communications},
  year = {2026},
  month = jun,
  doi = {10.1038/s41467-026-73846-y},
  url = {https://doi.org/10.1038/s41467-026-73846-y},
  note = {CC BY},
}
RIS
TY  - JOUR
TI  - End-to-end multimodal structure elucidation from raw spectra combining contrastive learning and evolutionary algorithms
AU  - Adrian Mirza
AU  - Luc Patiny
AU  - Kevin Maik Jablonka
PY  - 2026
DA  - 2026/06/05
JO  - Nature Communications
DO  - 10.1038/s41467-026-73846-y
UR  - https://doi.org/10.1038/s41467-026-73846-y
AB  - Elucidating molecular structures from spectroscopic data remains one of chemistry's most fundamental challenges, typically requiring extensive expert knowledge and manual interpretation of multiple analytical techniques. This is because the structure elucidation problem often has degenerate solutions for a limited set of experimental data. Existing computational approaches are limited to single spectroscopic modalities, require extensive manual preprocessing, and lack the confidence estimates and context necessary for practical application. Here we present SECS, a framework that combines contrastive learning with evolutionary algorithms to automate structure elucidation directly from raw, multimodal spectroscopic data. By aligning embeddings across NMR, infrared, and mass spectrometry, SECS mimics how experts use multiple spectroscopic lenses while providing calibrated confidence scores and relevant database context. On challenging molecular identification tasks, SECS matches expert chemist performance in head-to-head comparisons in a pilot study. The system successfully identifies incorrect structure assignments in published literature and adapts to new chemical domains without retraining by updating its reference database. Our approach demonstrates how synergistic combination of machine learning paradigms can solve analytical bottlenecks that have constrained chemical discovery.
SN  - 2041-1723
ER  -