Papers › MassSpecGym: A benchmark for the discovery and identification of molecules

MassSpecGym: A benchmark for the discovery and identification of molecules

30 Oct 2024arXiv:2410.23326archive 2025-07-28

Roman Bushuiev, Anton Bushuiev, Niek F. de Jonge, Adamo Young, Fleming Kretschmer, Raman Samusevich, Janne Heirman, Fei Wang, Luke Zhang, Kai Dührkop, Marcus Ludwig, Nils A. Haupt, Apurva Kalia, Corinna Brungs, Robin Schmid, Russell Greiner, Bo wang, David S. Wishart, Li-Ping Liu, Juho Rousu, Wout Bittremieux, Hannes Rost, Tytus D. Mak, Soha Hassoun, Florian Huber, Justin J. J. van der Hooft, Michael A. Stravs, Sebastian Böcker, Josef Sivic, Tomáš Pluskal

The discovery and identification of molecules in biological and environmental samples is crucial for advancing biomedical and chemical sciences. Tandem mass spectrometry (MS/MS) is the leading technique for high-throughput elucidation of molecular structures. However, decoding a molecular structure from its mass spectrum is exceptionally challenging, even when performed by human experts. As a result, the vast majority of acquired MS/MS spectra remain uninterpreted, thereby limiting our understanding of the underlying (bio)chemical processes. Despite decades of progress in machine learning applications for predicting molecular structures from MS/MS spectra, the development of new methods is severely hindered by the lack of standard datasets and evaluation protocols. To address this problem, we propose MassSpecGym -- the first comprehensive benchmark for the discovery and identification of molecules from MS/MS data. Our benchmark comprises the largest publicly available collection of high-quality labeled MS/MS spectra and defines three MS/MS annotation challenges: de novo molecular structure generation, molecule retrieval, and spectrum simulation. It includes new evaluation metrics and a generalization-demanding data split, therefore standardizing the MS/MS annotation tasks and rendering the problem accessible to the broad machine learning community. MassSpecGym is publicly available at https://github.com/pluskal-lab/MassSpecGym.

PaperPDFCodeCode Syntology ran

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, Syntology's MCP tool lists every function and class Syntology harvested from this paper and whether it ran (how to connect): get_harvested_code_for_paper(arxiv_id="2410.23326")

Code

Syntology Ran 2 of 6 code samples harvested from 1 repository linked to this paper; 4 have no recorded run. Of those that ran: 2 ran with no contract checked.

By repository: official repository: 6 samples from 1 repository, 2 ran. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

pluskal-lab/massspecgym officialmentioned in papermentioned on GitHubpytorchMIT report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

6 samples harvested; 2 ran; 0 honoured the contract we drafted; 4 have no recorded run. Read from Syntology's graph 2026-09-24; that is when this build read the record, not when the samples ran.

2ran
4unverified

Licence: 0 of the 6 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from pluskal-lab/massspecgym. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself; “our draft was wrong” and “fixture could not drive it” are failures of Syntology's instrument, not of the code.

Each sample ends with its code_sha256, Syntology's identity for that exact code. An agent fetches the stored sample with Syntology's MCP tool get_code(code_sha256="…") (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples from a repository marked official are listed first. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of four reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; or the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar). Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

register_atom_feat pluskal-lab/massspecgym/massspecgym/feat_utils.py official repository ran MIT (permissive) · 214108930a8afe6a · report
register_bond_feat pluskal-lab/massspecgym/massspecgym/feat_utils.py official repository ran MIT (permissive) · d43910dd9009ba2c · report
get_mol_feats_sizes pluskal-lab/massspecgym/massspecgym/feat_utils.py official repository unverified MIT (permissive) · 777d58c0554795e2 · report
load_massspecgym pluskal-lab/massspecgym/massspecgym/utils.py official repository unverified MIT (permissive) · 0072aa71818db60b · report
load_massspecgym_mols pluskal-lab/massspecgym/massspecgym/utils.py official repository unverified MIT (permissive) · e0cc16154f5452cf · report
load_unlabeled_mols pluskal-lab/massspecgym/massspecgym/utils.py official repository unverified MIT (permissive) · 2e2fbc4866fab30b · report

Tasks

De novo molecule generation from MS/MS spectrumDe novo molecule generation from MS/MS spectrum (bonus chemical formulae)MS/MS spectrum simulationMS/MS spectrum simulation (bonus chemical formulae)Molecule retrieval from MS/MS spectrumMolecule retrieval from MS/MS spectrum (bonus chemical formulae)

Datasets

Introduced by this paper, per the archive.

MassSpecGym

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
De novo molecule generation from MS/MS spectrum MassSpecGym Random chemical generation Top-1 Accuracy 0.00 #1 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym Random chemical generation Top-1 MCES 28.59 #1 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym Random chemical generation Top-1 Tanimoto 0.07 #1 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym Random chemical generation Top-10 Accuracy 0.00 #1 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym Random chemical generation Top-10 MCES 25.72 #1 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym Random chemical generation Top-10 Tanimoto 0.10 #1 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SELFIES Transformer Top-1 Accuracy 0.00 #2 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SELFIES Transformer Top-1 MCES 33.28 #2 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SELFIES Transformer Top-1 Tanimoto 0.10 #2 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SELFIES Transformer Top-10 Accuracy 0.00 #2 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SELFIES Transformer Top-10 MCES 21.84 #2 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SELFIES Transformer Top-10 Tanimoto 0.15 #2 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SMILES Transformer Top-1 Accuracy 0.00 #3 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SMILES Transformer Top-1 MCES 53.80 #3 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SMILES Transformer Top-1 Tanimoto 0.07 #3 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SMILES Transformer Top-10 Accuracy 0.00 #3 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SMILES Transformer Top-10 MCES 21.97 #3 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum MassSpecGym SMILES Transformer Top-10 Tanimoto 0.17 #3 of 3 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random chemical generation Top-1 Accuracy 0.00 #3 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random chemical generation Top-1 MCES 21.11 #3 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random chemical generation Top-1 Tanimoto 0.08 #3 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random chemical generation Top-10 Accuracy 0.00 #3 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random chemical generation Top-10 MCES 18.25 #3 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random chemical generation Top-10 Tanimoto 0.11 #3 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SELFIES Transformer Top-1 Accuracy 0.00 #6 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SELFIES Transformer Top-1 MCES 38.88 #6 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SELFIES Transformer Top-1 Tanimoto 0.08 #6 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SELFIES Transformer Top-10 Accuracy 0.00 #6 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SELFIES Transformer Top-10 MCES 26.87 #6 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SELFIES Transformer Top-10 Tanimoto 0.13 #6 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SMILES Transformer Top-1 Accuracy 0.00 #9 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SMILES Transformer Top-1 MCES 79.39 #9 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SMILES Transformer Top-1 Tanimoto 0.03 #9 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SMILES Transformer Top-10 Accuracy 0.00 #9 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SMILES Transformer Top-10 MCES 52.13 #9 of 9 Archive leaderboard report
De novo molecule generation from MS/MS spectrum (bonus chemical formulae) MassSpecGym SMILES Transformer Top-10 Tanimoto 0.10 #9 of 9 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FraGNNet Cosine Similarity 0.52 #1 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FraGNNet Hit Rate @ 1 46.64 #1 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FraGNNet Hit Rate @ 20 83.58 #1 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FraGNNet Hit Rate @ 5 72.56 #1 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FraGNNet Jensen-Shannon Similarity 0.47 #1 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FFN Fingerprint Cosine Similarity 0.25 #2 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FFN Fingerprint Hit Rate @ 1 8.44 #2 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FFN Fingerprint Hit Rate @ 20 38.57 #2 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FFN Fingerprint Hit Rate @ 5 21.43 #2 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym FFN Fingerprint Jensen-Shannon Similarity 0.24 #2 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym GNN Cosine Similarity 0.19 #3 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym GNN Hit Rate @ 1 3.95 #3 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym GNN Hit Rate @ 20 26.27 #3 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym GNN Hit Rate @ 5 11.92 #3 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym GNN Jensen-Shannon Similarity 0.20 #3 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym Precursor m/z Cosine Similarity 0.15 #4 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym Precursor m/z Hit Rate @ 1 0.38 #4 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym Precursor m/z Hit Rate @ 20 7.17 #4 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym Precursor m/z Hit Rate @ 5 1.72 #4 of 4 Archive leaderboard report
MS/MS spectrum simulation MassSpecGym Precursor m/z Jensen-Shannon Similarity 0.15 #4 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym FraGNNet Hit Rate @ 1 31.93 #1 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym FraGNNet Hit Rate @ 20 82.70 #1 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym FraGNNet Hit Rate @ 5 63.20 #1 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym FFN Fingerprint Hit Rate @ 1 7.62 #2 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym FFN Fingerprint Hit Rate @ 20 44.12 #2 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym FFN Fingerprint Hit Rate @ 5 22.70 #2 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym GNN Hit Rate @ 1 3.63 #3 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym GNN Hit Rate @ 20 33.77 #3 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym GNN Hit Rate @ 5 13.55 #3 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym Precursor m/z Hit Rate @ 1 2.09 #4 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym Precursor m/z Hit Rate @ 20 22.65 #4 of 4 Archive leaderboard report
MS/MS spectrum simulation (bonus chemical formulae) MassSpecGym Precursor m/z Hit Rate @ 5 8.52 #4 of 4 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym MIST Hit rate @ 1 14.64 #3 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym MIST Hit rate @ 20 59.15 #3 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym MIST Hit rate @ 5 34.87 #3 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym MIST MCES @ 1 15.37 #3 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym DeepSets + Fourier features Hit rate @ 1 5.24 #5 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym DeepSets + Fourier features Hit rate @ 20 28.21 #5 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym DeepSets + Fourier features Hit rate @ 5 12.58 #5 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym DeepSets + Fourier features MCES @ 1 22.13 #5 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym Fingerprint FFN Hit rate @ 1 2.54 #6 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym Fingerprint FFN Hit rate @ 20 20.00 #6 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym Fingerprint FFN Hit rate @ 5 7.59 #6 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym Fingerprint FFN MCES @ 1 24.66 #6 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym DeepSets Hit rate @ 1 1.47 #7 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym DeepSets Hit rate @ 20 19.23 #7 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym DeepSets Hit rate @ 5 6.21 #7 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym DeepSets MCES @ 1 25.11 #7 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym Random Hit rate @ 1 0.37 #8 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym Random Hit rate @ 20 8.22 #8 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym Random Hit rate @ 5 2.01 #8 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum MassSpecGym Random MCES @ 1 30.81 #8 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym MIST Hit rate @ 1 9.57 #4 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym MIST Hit rate @ 20 41.12 #4 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym MIST Hit rate @ 5 22.11 #4 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym MIST MCES @ 1 12.75 #4 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym DeepSets + Fourier features Hit rate @ 1 6.56 #5 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym DeepSets + Fourier features Hit rate @ 20 33.46 #5 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym DeepSets + Fourier features Hit rate @ 5 16.46 #5 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym DeepSets + Fourier features MCES @ 1 14.14 #5 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym Fingerprint FFN Hit rate @ 1 5.09 #6 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym Fingerprint FFN Hit rate @ 20 31.97 #6 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym Fingerprint FFN Hit rate @ 5 14.69 #6 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym Fingerprint FFN MCES @ 1 14.94 #6 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym DeepSets Hit rate @ 1 4.42 #7 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym DeepSets Hit rate @ 20 30.76 #7 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym DeepSets Hit rate @ 5 14.46 #7 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym DeepSets MCES @ 1 15.04 #7 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random Hit rate @ 1 3.06 #8 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random Hit rate @ 20 27.74 #8 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random Hit rate @ 5 11.35 #8 of 8 Archive leaderboard report
Molecule retrieval from MS/MS spectrum (bonus chemical formulae) MassSpecGym Random MCES @ 1 13.87 #8 of 8 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections