Fragment-Grounded Neural Simulation of Electron Ionization Mass Spectra at Library Scale
Abstract
Electron ionization mass spectrometry (EI-MS) is a sensitive, reproducible analytical technique, yet compound identification conventionally relies on reference databases that capture only a fraction of accessible chemical space. In silico prediction of EI-MS spectra can extend these libraries, but existing data-driven models generalize poorly beyond their training distribution, and physics-based simulations are too costly for library-scale use. Here, we present a fragment-grounded neural workflow that predicts an EI-MS spectrum by exhaustively enumerating the fragmentation graph of a molecule and then assigning intensities to the resulting substructures. By relating every predicted peak back to a candidate substructure, our model generalizes more reliably to unseen scaffolds and out-of-distribution chemistry than prior fingerprint-and black-box transformer-based predictors. On a scaffold split, it improves top-1 retrieval accuracy from formula-matched compounds by 32% relative to the strongest baseline, and it matches the accuracy of quantum-chemical simulation while being roughly seven orders of magnitude faster. To test the model’s capabilities under a harder, formula-agnostic setting, we further evaluate retrieval against a library of 87.3 million compounds from
Related
- Mechanism-grounded fragmentation trees enable explicit chemical reasoning for electron ionization mass spectrum prediction
- Electron spectra from strong-field enhanced ionization in heavy water
- Tree-Based Classification of Functional Groups from IR Spectra: Feature Importance Corroborates Quantum-Chemical Fragment Frequencies
- Ionization Energies, Electron Affinities, Bandgaps, Exciton Binding Energies, and Polarization Energies of Orientation-Controlled Picene, [6]-Phenacene, and [7]-Phenacene Thin Films
- Vertical ionization energies from inverse generalized Kohn-Sham calculations at fixed electron density
BibTeX
@misc{lederbauer2026fragmentgrounded,
title = {{Fragment-Grounded Neural Simulation of Electron Ionization Mass Spectra at Library Scale}},
author = {Magdalena Lederbauer and Runzhong Wang and Connor W. Coley},
howpublished = {ChemRxiv},
year = {2026},
month = sep,
doi = {10.26434/chemrxiv.15008135/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008135/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Fragment-Grounded Neural Simulation of Electron Ionization Mass Spectra at Library Scale
AU - Magdalena Lederbauer
AU - Runzhong Wang
AU - Connor W. Coley
PY - 2026
DA - 2026/09/01
PB - ChemRxiv
DO - 10.26434/chemrxiv.15008135/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008135/v1
AB - Electron ionization mass spectrometry (EI-MS) is a sensitive, reproducible analytical technique, yet compound identification conventionally relies on reference databases that capture only a fraction of accessible chemical space. In silico prediction of EI-MS spectra can extend these libraries, but existing data-driven models generalize poorly beyond their training distribution, and physics-based simulations are too costly for library-scale use. Here, we present a fragment-grounded neural workflow that predicts an EI-MS spectrum by exhaustively enumerating the fragmentation graph of a molecule and then assigning intensities to the resulting substructures. By relating every predicted peak back to a candidate substructure, our model generalizes more reliably to unseen scaffolds and out-of-distribution chemistry than prior fingerprint-and black-box transformer-based predictors. On a scaffold split, it improves top-1 retrieval accuracy from formula-matched compounds by 32% relative to the strongest baseline, and it matches the accuracy of quantum-chemical simulation while being roughly seven orders of magnitude faster. To test the model’s capabilities under a harder, formula-agnostic setting, we further evaluate retrieval against a library of 87.3 million compounds from
ER -