What this source contributed
Records in this release that cite this source. A record can cite several sources.
Results
Values as the source printed them. Compare results only under compatible protocols, datasets and metrics.
Claims
Statements this source supports, each with where in the source it was checked.
- Baseline: Shortlist molecular identities from tandem mass spectra: Baseline MassSpecGym DeepSets fingerprint baseline (part2-massspecgym-arxiv-v1: Section 4.1 (Random and DeepSets retrieval baselines), Supplementary (DeepSets architecture) and Table 3; uc20260930-source-msalign-v2: results tables (DeepSets rows on the MassSpecGym and SpectraVerse formula-free protocols))
- DeepSet [MSAlign v2] family identity: Configuration of MassSpecGym DeepSets fingerprint predictor (Table 3 source column DeepSet; Section 5.1 Baselines)
- DeepSet [MSAlign v2]: facets.method_types: Facets: method types ["supervised_machine_learning"] (MassSpecGym Section 4.1: Fingerprint FFN "employs a feedforward neural network to predict the Morgan fingerprint", and DeepSets; MSAlign arXiv 2605.19752 Table 1 and "Baselines" (v1 and v2): JESTR "pretrains with batch InfoNCE", Emb-Cos "uses candidates InfoNCE to learn a shared space", SAIL is trained with the batch loss)
- DeepSets: facets.method_types: Facets: method types ["supervised_machine_learning"] (MassSpecGym Section 4.1: Fingerprint FFN "employs a feedforward neural network to predict the Morgan fingerprint", and DeepSets; MSAlign arXiv 2605.19752 Table 1 and "Baselines" (v1 and v2): JESTR "pretrains with batch InfoNCE", Emb-Cos "uses candidates InfoNCE to learn a shared space", SAIL is trained with the batch loss)
- Emb-Cos [MSAlign v2] family identity: Configuration of Emb-Cos (Table 3 source column Emb-Cos; Section 5.1 Baselines)
- Emb-Cos [MSAlign v2]: facets.method_types: Facets: method types ["supervised_machine_learning"] (MassSpecGym Section 4.1: Fingerprint FFN "employs a feedforward neural network to predict the Morgan fingerprint", and DeepSets; MSAlign arXiv 2605.19752 Table 1 and "Baselines" (v1 and v2): JESTR "pretrains with batch InfoNCE", Emb-Cos "uses candidates InfoNCE to learn a shared space", SAIL is trained with the batch loss)
- Emb-Cos: description: Description Contrastive model that learns a shared embedding space for molecules and MS/MS spectra with candidate InfoNCE to retrieve a metabolite's structure from its spectrum among candidate molecules, without pretrained encoders or score fusion. (uc20260930-source-msalign-v2: Section 5.1 'Baselines'; Table 3 caption)
- Emb-Cos: facets.method_types: Facets: method types ["supervised_machine_learning"] (MassSpecGym Section 4.1: Fingerprint FFN "employs a feedforward neural network to predict the Morgan fingerprint", and DeepSets; MSAlign arXiv 2605.19752 Table 1 and "Baselines" (v1 and v2): JESTR "pretrains with batch InfoNCE", Emb-Cos "uses candidates InfoNCE to learn a shared space", SAIL is trained with the batch loss)
- Evidence summary: Shortlist molecular identities from tandem mass spectra: Summary Two preprints report formula-free retrieval of the true structure from mass-matched PubChem candidates for MassSpecGym spectra, and their values are not directly comparable. In the MSAlign preprint, MSAlign ranks first among the formula-free methods on every split the paper reports, but its top-1 recall falls from about half on the formula split to under a fifth on the MCES split. On the MassSpecGym formula split, Recall@1 was 47.5 for MSAlign and 58.2 with score fusion, against 42.7 for Emb-Cos, 13.9 for JESTR and 4.7 for DeepSet, with standard deviations of 0.5 to 3.4 over three random splitting seeds. On the MCES split the order is the same but the values are much lower: 19.0, 30.8, 12.6, 7.9 and 1.9; the table caption describes three splits there too, but the methods describe two swapped variants. Recall@20 reaches 87.0 for MSAlign and 92.8 with score fusion on the formula split, and 64.9 and 82.0 on MCES. An earlier version of the same paper reports Recall@1 of 53.8 for MSAlign and 41.2 for Emb-Cos on the formula split, so the two versions differ and are not independent measurements. These MSAlign values assume the true structure is among 256 mass-matched candidates and were scored by the MSAlign authors. MassSpecGym in the Wild, whose authors include the developers of MIST, MIST-CF and MassSpecGym, reports two formula-free configurations that it does not flag for formula leakage, on the MassSpecGym test fold with up to 256 mass-matched candidates per spectrum. MIST given the top formula predicted by MIST-CF found the true structure at rank 1 for 26.64 percent of test spectra and within the top 20 for 54.90 percent, against 0.43 and 7.99 percent for random ranking. Its candidate pool and split differ from MSAlign's, and the MIST score depends on the formula annotator as well as the retriever. All of these values are reported by the authors of preprints, and none measures open-world identification, because the true structure is always among the candidates. (uc20260930-source-msalign-v2: Table 3 p7, massspecgym-formula, massspecgym-mces and spectraverse blocks, R@1 and R@20 rows; evidence-official-760ab2fa8c396aeb796c: Table 3, no-formula blocks, R@1 and R@20 rows; lit-otheromics-20261011-msgwild-source-liu2026: Table 2, rows 'Random' and 'MIST (MIST-CF top-1 formula)', columns 'Hit rate@1' and 'Hit rate@20'; Section 4.3)
- FFN: facets.method_types: Facets: method types ["supervised_machine_learning"] (MassSpecGym Section 4.1: Fingerprint FFN "employs a feedforward neural network to predict the Morgan fingerprint", and DeepSets; MSAlign arXiv 2605.19752 Table 1 and "Baselines" (v1 and v2): JESTR "pretrains with batch InfoNCE", Emb-Cos "uses candidates InfoNCE to learn a shared space", SAIL is trained with the batch loss)
- FLARE [MSAlign v2] family identity: Configuration of FLARE (Table 3 source column FLARE; Section 5.1 Baselines)
- FLARE: description: Description Contrastive spectrum-to-molecule retrieval model that follows MVP with a stronger spectra encoder using subformula peak annotation and a richer similarity; it needs the target's molecular formula. (uc20260930-source-msalign-v2: Section 5.1 'Baselines'; Table 3 caption)
- JESTR [MSAlign v2] family identity: Configuration of JESTR (Table 3 source column JESTR; Section 5.1 Baselines)
- JESTR [MSAlign v2]: facets.method_types: Facets: method types ["supervised_machine_learning"] (MassSpecGym Section 4.1: Fingerprint FFN "employs a feedforward neural network to predict the Morgan fingerprint", and DeepSets; MSAlign arXiv 2605.19752 Table 1 and "Baselines" (v1 and v2): JESTR "pretrains with batch InfoNCE", Emb-Cos "uses candidates InfoNCE to learn a shared space", SAIL is trained with the batch loss)
- JESTR: description: Description Contrastive model that matches MS/MS spectra to candidate molecules for metabolite retrieval, pretrained with batch InfoNCE and fine-tuned with candidate InfoNCE. (uc20260930-source-msalign-v2: Section 5.1 'Baselines'; Table 3 caption)
- JESTR: facets.method_types: Facets: method types ["supervised_machine_learning"] (MassSpecGym Section 4.1: Fingerprint FFN "employs a feedforward neural network to predict the Morgan fingerprint", and DeepSets; MSAlign arXiv 2605.19752 Table 1 and "Baselines" (v1 and v2): JESTR "pretrains with batch InfoNCE", Emb-Cos "uses candidates InfoNCE to learn a shared space", SAIL is trained with the batch loss)
- MassSpecGym DeepSets fingerprint predictor: description: Description DeepSets fingerprint prediction baseline from MassSpecGym that predicts a molecular fingerprint from an MS/MS spectrum to rank candidate molecules (Fourier-feature variant in MSAlign). (uc20260930-source-msalign-v2: Section 5.1 'Baselines'; Table 3 caption)
- MIST [MSAlign v2] family identity: Configuration of MIST (Table 3 source column MIST; Section 5.1 Baselines)
- MIST: description: Description Spectrum-to-molecule retrieval model that pretrains on fingerprint prediction, fine-tunes with contrastive learning and augments its data with a forward model that simulates spectra; it needs the target's molecular formula. (uc20260930-source-msalign-v2: Section 5.1 'Baselines'; Table 3 caption)
- MSAlign [MSAlign v2] family identity: Configuration of MSAlign (Table 3 source column MSAlign; Section 5.1 Baselines)
- MSAlign [MSAlign v2]: facets.method_types: Facets: method types ["foundation_model"] (MSAlign arXiv 2605.19752v2 abstract and Section 3 "Foundation model alignment" (p.4): aligns "two frozen foundation models (DreaMS for mass spectra and MolDeBERTa for molecules)" through trained MLP adapters)
- MSAlign score fusion (overbar) [MSAlign v2] family identity: Configuration of MSAlign (Table 3 source column MSAlign score fusion (overbar); Section 5.1 Baselines)
- MSAlign score fusion (overbar) [MSAlign v2]: facets.method_types: Facets: method types ["foundation_model"] (MSAlign arXiv 2605.19752v2 abstract and Section 3 "Foundation model alignment" (p.4): aligns "two frozen foundation models (DreaMS for mass spectra and MolDeBERTa for molecules)" through trained MLP adapters)
- MSAlign score fusion +Filter (overbar) [MSAlign v2] family identity: Configuration of MSAlign (Table 3 source column MSAlign score fusion +Filter (overbar); Section 5.1 Baselines)
- MSAlign v2 retrieval: massspecgym-formula-formula-free study membership: Part of MSAlign molecular retrieval evaluations (MSAlign 2605.19752v2 Table 3; Section 5.1)
Models, benchmarks and other records
Models (6)
Configurations (9)
Protocols (6)
- MSAlign v2 retrieval: massspecgym-formula-formula-free
- MSAlign v2 retrieval: massspecgym-formula-formula-oracle
- MSAlign v2 retrieval: massspecgym-mces-formula-free
- MSAlign v2 retrieval: massspecgym-mces-formula-oracle
- MSAlign v2 retrieval: spectraverse-formula-free
- MSAlign v2 retrieval: spectraverse-formula-oracle
Dataset subsets (6)
Baselines (3)
Evaluations (27)
- DeepSet massspecgym-formula-formula-free
- DeepSet massspecgym-mces-formula-free
- DeepSet spectraverse-formula-free
- Emb-Cos massspecgym-formula-formula-free
- Emb-Cos massspecgym-mces-formula-free
- Emb-Cos spectraverse-formula-free
- FLARE massspecgym-formula-formula-oracle
- FLARE massspecgym-mces-formula-oracle
- FLARE spectraverse-formula-oracle
- JESTR massspecgym-formula-formula-free
- JESTR massspecgym-mces-formula-free
- JESTR spectraverse-formula-free
- MIST massspecgym-formula-formula-oracle
- MIST massspecgym-mces-formula-oracle
- MIST spectraverse-formula-oracle
- MSAlign massspecgym-formula-formula-free
- MSAlign massspecgym-mces-formula-free
- MSAlign score fusion (overbar) massspecgym-formula-formula-free
- MSAlign score fusion (overbar) massspecgym-mces-formula-free
- MSAlign score fusion (overbar) spectraverse-formula-free
- MSAlign score fusion +Filter (overbar) massspecgym-formula-formula-oracle
- MSAlign score fusion +Filter (overbar) massspecgym-mces-formula-oracle
- MSAlign score fusion +Filter (overbar) spectraverse-formula-oracle
- MSAlign spectraverse-formula-free
- MVP massspecgym-formula-formula-oracle
Evidence
Source checking verifies the cited claim or transcription. It does not establish independent reproduction.
Evidence table
Inspect claims, sources and review details
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
16 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Artifact SHA-256 a31d1167b0ded72f64717881dd6cba0946a56b495d76f169d8e44115c3266fda Source metadata | MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Artifact url https://arxiv.org/pdf/2605.19752v2 Source metadata | MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Retrieval note Unversioned URL returned PDF explicitly labelled 2605.19752v2. Prior v1 catalogue source hash differs; old scientific records preserved. Source metadata | MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Retrieval url https://arxiv.org/pdf/2605.19752 Source metadata | MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Retrieved at 2026-09-30 Source metadata | MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Review method1 values
| MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Review: scope Only explicitly extracted table rows and protocol methods; no execution. Source metadata | MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Url https://arxiv.org/pdf/2605.19752v2 Source metadata | MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Version 2605.19752v2, 25 September 2026 Source metadata | MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Description Primary source retrieved and inspected for this audit. Source metadata | MSAlign arXiv v2 Table 3 No field-specific location recorded Version: 2605.19752v2, 25 September 2026 | catalogued No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Sources and history
Release 2026-11-21-c8bea25e5abb · Record review: source checked
Primary source retrieved and inspected for this audit.
Technical metadata and extraction receipts
Stable ID: uc20260930-source-msalign-v2
- url
- https://arxiv.org/pdf/2605.19752v2
- artifact url
- https://arxiv.org/pdf/2605.19752v2
- retrieved at
- 2026-09-30
- version
- 2605.19752v2, 25 September 2026
- artifact sha256
- a31d1167b0ded72f64717881dd6cba0946a56b495d76f169d8e44115c3266fda
- retrieval url
- https://arxiv.org/pdf/2605.19752
- retrieval note
- Unversioned URL returned PDF explicitly labelled 2605.19752v2. Prior v1 catalogue source hash differs; old scientific records preserved.
- review
- method: automated-source-review; scope: Only explicitly extracted table rows and protocol methods; no execution.