On the NPLIB1 benchmark, MARLIN is the strongest method evaluated without a ground-truth formula across exact-match accuracy, structural distance, and fingerprint similarity, and it recovers the correct molecular formula as a byproduct about as often as a dedicated predictor without ever using one.
Abstract
Untargeted tandem mass spectrometry (MS/MS) detects thousands of small molecules per biological sample, yet most go unidentified because they are absent from spectral libraries. These uncharacterized metabolites and natural products are precisely the compounds that matter for drug discovery, biomarker research, and exposomics. Computational de novo structure elucidation could close this gap, but almost all state-of-the-art methods assume the ground-truth molecular formula is known, an oracle that does not exist for genuinely novel compounds and is itself predicted with substantial error. We present MARLIN, a de novo method that elucidates structures directly from a spectrum with no molecular formula at any stage. A self-supervised encoder predicts a molecular fingerprint from the raw peaks, and a block-diffusion language model generates candidate structures conditioned only on the fingerprint and the instrument-measured precursor mass. A provably safe mass-shell constraint keeps every candidate consistent with the measured mass without fixing the atom inventory, and candidates are accepted by exact parts-per-million mass agreement. A symmetric noise objective absorbs encoder error, and a candidate-diversity mechanism keeps the candidates from collapsing to a single structure. On the NPLIB1 benchmark, MARLIN is the strongest method evaluated without a ground-truth formula across exact-match accuracy, structural distance, and fingerprint similarity, and it recovers the correct molecular formula as a byproduct about as often as a dedicated predictor without ever using one. MARLIN enables reliable de novo structure elucidation in the realistic discovery regime where the molecular formula is unavailable.
Compound identification and structure annotation from mass spectra are essential in drug detection, forensics, and small molecule discovery. Current approaches to compound identification from electron ionization mass spectra (EI-MS) are dependent on different forms of searching databases that are orders of magnitude smaller than the space of potential molecular structures they attempt to cover. We introduce SpecTUS: Spectral Translator for Unknown Structures, a deep learning model for de novo structural annotation, translating gas chromatography EI-MS spectra directly into molecular structures without requiring reference databases. This enables the identification of novel compounds absent from spectral libraries. In a rigorous evaluation, SpecTUS significantly outperformed standard database search techniques. On a held-out test set of 28,267 spectra from NIST 20, the model’s single suggestion perfectly reconstructed 43% of the subset’s compounds. On 76% of this test set, the single suggestion is strictly better, in terms of Tanimoto similarity of Morgan fingerprint, than the result of hybrid database search. With ten suggestions, SpecTUS achieved 65% perfect reconstructions, surpassing hybrid search on 84% of the test set.
Adam Hájek, Michal Starý, E. Price et al.· Analytical Chemistry· 0 citations
Molecular structure elucidation from tandem mass spectra (MS/MS) is a central inverse problem in analytical chemistry. Most existing approaches to MS/MS identification remain tied to reference libraries or predefined candidate sets, whereas de novo methods aim to generate structures directly from spectra. A common de novo route predicts a molecular fingerprint from the spectrum and then decodes structures from it, enabling decoder pretraining on large molecule-only corpora. However, this paradigm creates a training-inference mismatch: the decoder is trained on oracle fingerprints computed from molecules, but at inference it is queried with a noisy spectrum-induced fingerprint posterior that is typically collapsed to a single thresholded fingerprint. We introduce MS-GPT, which recasts fingerprint-mediated de novo elucidation as spectrum-induced posterior querying of a conditional molecule-language model. MS-GPT conditions a molecule-language model on fingerprints and formulas, then converts the spectrum-induced posterior into a band of fingerprint queries near the oracle-fingerprint manifold through active-bit density calibration. Candidates sampled across this band are pooled and ranked by generation-frequency consensus. A lightweight LoRA adapter further mitigates domain-specific posterior bias while preserving the pretrained molecular prior. On NPLIB1 and MassSpecGym, MS-GPT sets a new state of the art, reaching Top-1/Top-10 exact-match accuracy of 29.8\%/41.1\% and 23.9\%/28.7\%, respectively. Candidate-pool scaling shows that efficient autoregressive molecular generation continues to improve recall with a little additional inference cost. The source code and model checkpoints are available at https://github.com/VIKI623/MS-GPT.
Xin Zhao, Yumin Liu, Zhuowei Li et al.· 0 citations
AIMe (AI Molecule Explorer), a multi-agent neuro-symbolic AI framework that transforms the interpretation of unknown spectra into an omics-scale exploration across the known structural space, providing chemically interpretable annotations, is introduced.
U. Acikalin, Dieqiao Feng, Aaron M. Ferber et al.· bioRxiv· 0 citations
Untargeted metabolomics often results in a significant portion of unannotated metabolites, or “metabolic dark matter,” which hinders biological interpretation. A two-step analytical approach was developed to systematically prioritize and interpret unannotated metabolites using plasma LC–MS/MS data from pregnant women with obesity as a biologically relevant test dataset. The first step involved clustering 1,021 known metabolites into ten structurally coherent groups based on the Tanimoto similarity, thus defining the biologically relevant chemical space of the dataset. These metabolites were further characterized by Absorption, Distribution, Metabolism, and Excretion (ADME) profiling, protein target prediction, molecular docking and Kyoto Encyclopedia of Genes and Genomes pathway mapping analysis, to establish biological plausibility and functional perspective. Candidate structures for 1,836 unannotated features were retrieved from PubChem using molecular formula and molecular weight matching within a ±0.5 Da tolerance. This search yielded 569,115 candidate structures, of which 368,197 unique structures were retained after curation. Tanimoto coefficient filtering reduced the candidate pool to 19,868 structurally plausible candidates, and retention time-based prioritization further refined this set to 418 high confidence candidate annotations, including 83 database-supported candidates identified through HMDB and LIPID MAPS structure database cross-referencing. RT-based prioritization effectively distinguished positional isomers sharing the same molecular formula by incorporating agreement between predicted and experimentally observed retention times. This improved discrimination among structurally similar candidates, expanded metabolite annotation confidence, and provided a scalable framework for prioritizing dark matter metabolites in untargeted metabolomics. Clustered-based workflow integrating chemical similarity and retention time to prioritize and annotate unknown metabolites
D. Bhandari, H. Paz, Keith Henderson et al.· Metabolomics· 0 citations
Determining molecular structures from spectroscopic data remains fundamentally challenging because the inverse problem is intrinsically underdetermined: individual spectra are sparse, low-dimensional, and encode only partial structural evidence relative to the vast space of possible molecules. We address this challenge by formulating automated structure elucidation as a scalable hypothesis-refinement paradigm that tightly integrates spectral evidence with large-scale molecular priors. To supply structure-resolving NMR signals for multimodal learning, we construct \textbf{QM9SPIN}, a DFT-derived dataset comprising diverse 1D and 2D spectra, including J-coupling, DEPT experiments, and explicit spin--spin interactions. On this foundation, we introduce \textbf{SpectroMol}, a spectrum-to-structure model that proposes chemically valid molecular hypotheses conditioned on multimodal spectral inputs. Complementarily, we develop \textbf{MS-Mol2Mol}, a high-resolution mass-constrained molecular generator that integrates molecular formula, exact mass, and degree of unsaturation within a conditional generative prior trained on 400 million molecules, ensuring global compositional consistency and chemically realistic refinement. The integrated system achieves 93.8\% top-1 accuracy on the simulated benchmark, adapts effectively from simulated to experimental spectra with limited experimental fine-tuning, and further improves experimental predictions through mass-guided refinement, establishing a scalable route toward automated, data-driven organic structure elucidation.
Chengchun Liu, Zhiyuan Yan, Li Yuan et al.· 0 citations
Machine learning methods for predicting the electron ionization mass spectra from molecular structures have shown promise for environmental chemical identification, but their performance under domain-specific data scarcity remains poorly understood. We systematically compare a conventional multilayer perceptron model (NEIMS) with a Transformer-based chemical foundation model (MolFormer-XL) for the electron ionization mass spectrometry spectrum prediction under controlled few-shot conditions. Using fluorine-containing molecules as a broader proxy domain, including a PFAS-like subset, motivated by the practical challenge of detecting novel fluorinated contaminants with limited reference data, we vary the number of domain-specific training examples from 5 to 175 while maintaining fixed validation and test sets. Across all few-shot conditions and three of four evaluation metrics (weighted cosine similarity, intensity-weighted precision, and top-10 precision), MolFormer-XL consistently outperforms NEIMS, while intensity-weighted recall remains comparable between the two models. The largest performance gaps are observed in extreme data-scarcity regimes. These results demonstrate that MolFormer-XL, which combines pre-trained molecular representations with a Transformer-based architecture and learned SMILES embeddings, provides a promising approach for transfer under severe domain-specific data scarcity in environmental mass spectrometry.
