MassSpecGym · formula
Dataset and cohort used in the cited comparison. Dataset population counts do not establish successful prediction coverage.
Dataset and evaluation context
A dataset supplies biological observations. The evaluation protocol defines how those observations are split, used and scored.
Evaluation results
Release 2026-09-17-d277315f7d76 · 4 evaluations · 20 metric rows. Different protocols are not a single leaderboard.
| Metric and finding | Coverage and uncertainty | Evidence |
|---|---|---|
| Precursor m/z (formula): MassSpecGym · formula Configuration: Precursor m/z (formula)Protocol: MassSpecGym · formula (MassSpecGym Spectrum simulation)Dataset: MassSpecGym · formula Predict a spectrum from a molecule under measurement conditions. Metadata-complete simulation subset; simulation-derived retrieval is distinct from direct molecule retrieval in Table 3. MCES single-linkage molecular clustering at threshold 10; fixed held-out test split Author-reported evaluation · Evaluation metadata: needs review | ||
| 2.09 (1.66-2.59)% Hit rate @1 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 1.66; upper: 2.59 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, Precursor m/z, Hit rate @1 Source checking is not independent reproduction. |
| 8.52 (7.65-9.53)% Hit rate @5 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 7.65; upper: 9.53 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, Precursor m/z, Hit rate @5 Source checking is not independent reproduction. |
| - Jensen-Shannon similarity Unit: dimensionless · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, Precursor m/z, Jensen-Shannon similarity Source checking is not independent reproduction. |
| - Cosine similarity Unit: dimensionless · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, Precursor m/z, Cosine similarity Source checking is not independent reproduction. |
| 22.65 (21.26-24.01)% Hit rate @20 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 21.26; upper: 24.01 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, Precursor m/z, Hit rate @20 Source checking is not independent reproduction. |
| GNN (formula): MassSpecGym · formula Configuration: GNN (formula)Protocol: MassSpecGym · formula (MassSpecGym Spectrum simulation)Dataset: MassSpecGym · formula Predict a spectrum from a molecule under measurement conditions. Metadata-complete simulation subset; simulation-derived retrieval is distinct from direct molecule retrieval in Table 3. MCES single-linkage molecular clustering at threshold 10; fixed held-out test split Author-reported evaluation · Evaluation metadata: needs review | ||
| - Jensen-Shannon similarity Unit: dimensionless · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, GNN, Jensen-Shannon similarity Source checking is not independent reproduction. |
| 33.77 (32.26-35.37)% Hit rate @20 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 32.26; upper: 35.37 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, GNN, Hit rate @20 Source checking is not independent reproduction. |
| 3.63 (3.05-4.29)% Hit rate @1 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 3.05; upper: 4.29 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, GNN, Hit rate @1 Source checking is not independent reproduction. |
| - Cosine similarity Unit: dimensionless · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, GNN, Cosine similarity Source checking is not independent reproduction. |
| 13.55 (12.46-14.68)% Hit rate @5 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 12.46; upper: 14.68 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, GNN, Hit rate @5 Source checking is not independent reproduction. |
| FFN Fingerprint (formula): MassSpecGym · formula Configuration: FFN Fingerprint (formula)Protocol: MassSpecGym · formula (MassSpecGym Spectrum simulation)Dataset: MassSpecGym · formula Predict a spectrum from a molecule under measurement conditions. Metadata-complete simulation subset; simulation-derived retrieval is distinct from direct molecule retrieval in Table 3. MCES single-linkage molecular clustering at threshold 10; fixed held-out test split Author-reported evaluation · Evaluation metadata: needs review | ||
| - Jensen-Shannon similarity Unit: dimensionless · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FFN Fingerprint, Jensen-Shannon similarity Source checking is not independent reproduction. |
| 22.70 (21.32-24.12)% Hit rate @5 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 21.32; upper: 24.12 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FFN Fingerprint, Hit rate @5 Source checking is not independent reproduction. |
| 44.12 (42.51-45.75)% Hit rate @20 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 42.51; upper: 45.75 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FFN Fingerprint, Hit rate @20 Source checking is not independent reproduction. |
| - Cosine similarity Unit: dimensionless · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FFN Fingerprint, Cosine similarity Source checking is not independent reproduction. |
| 7.62 (6.77-8.54)% Hit rate @1 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 6.77; upper: 8.54 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FFN Fingerprint, Hit rate @1 Source checking is not independent reproduction. |
| FraGNNet (formula): MassSpecGym · formula Configuration: FraGNNet (formula)Protocol: MassSpecGym · formula (MassSpecGym Spectrum simulation)Dataset: MassSpecGym · formula Predict a spectrum from a molecule under measurement conditions. Metadata-complete simulation subset; simulation-derived retrieval is distinct from direct molecule retrieval in Table 3. MCES single-linkage molecular clustering at threshold 10; fixed held-out test split Author-reported evaluation · Evaluation metadata: needs review | ||
| - Cosine similarity Unit: dimensionless · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FraGNNet, Cosine similarity Source checking is not independent reproduction. |
| 31.93 (30.40-33.50)% Hit rate @1 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 30.40; upper: 33.50 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FraGNNet, Hit rate @1 Source checking is not independent reproduction. |
| 82.70 (81.45-83.93)% Hit rate @20 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 81.45; upper: 83.93 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FraGNNet, Hit rate @20 Source checking is not independent reproduction. |
| - Jensen-Shannon similarity Unit: dimensionless · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FraGNNet, Jensen-Shannon similarity Source checking is not independent reproduction. |
| 63.20 (61.64-64.76)% Hit rate @5 Unit: percent · Direction: higher | Uncertainty: type: bootstrap confidence interval; level: 0.999; resamples: 20000; lower: 61.64; upper: 64.76 Scored: Not reported · Eligible: Not reported | source checkedMassSpecGym: A benchmark for the discovery and identification of molecules · Table 4, bonus formula challenge, FraGNNet, Hit rate @5 Source checking is not independent reproduction. |
Evidence table
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.
6 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| attributes.reported_population No value recorded Context-only references | MassSpecGym: A benchmark for the discovery and identification of molecules Table 4: Cosine similarity, formula Version: 2410.23326v1 | missing or unspecified No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.source_locator Table 4: Cosine similarity, formula Context-only references | MassSpecGym: A benchmark for the discovery and identification of molecules Table 4: Cosine similarity, formula Version: 2410.23326v1 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.split MCES single-linkage molecular clustering at threshold 10; fixed held-out test split Context-only references | MassSpecGym: A benchmark for the discovery and identification of molecules Table 4: Cosine similarity, formula Version: 2410.23326v1 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.subset formula Context-only references | MassSpecGym: A benchmark for the discovery and identification of molecules Table 4: Cosine similarity, formula Version: 2410.23326v1 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| description Dataset and cohort used in the cited comparison. Dataset population counts do not establish successful prediction coverage. Context-only references | MassSpecGym: A benchmark for the discovery and identification of molecules Table 4: Cosine similarity, formula Version: 2410.23326v1 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| name MassSpecGym · formula Context-only references | MassSpecGym: A benchmark for the discovery and identification of molecules Table 4: Cosine similarity, formula Version: 2410.23326v1 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
Sources and history
Release 2026-09-17-d277315f7d76 · Record review: needs review
- MassSpecGym: A benchmark for the discovery and identification of molecules · Original source · 2410.23326v1
Technical metadata and extraction receipts
Stable ID: paper-dataset-712fb45862f6d8614f
- areas
- metabolomics
- split
- MCES single-linkage molecular clustering at threshold 10; fixed held-out test split
- subset
- formula
- reported population
- Not reported
- source locator
- Table 4: Cosine similarity, formula
- missing metadata
- manifest: unextracted; scored count: unreported
- entity classification
- review date: 2026-09-17; rationale: This record identifies a biological data collection or source-labelled evaluation cohort. Keep its dataset identity; split, assay, taxonomic level, candidate restrictions and comparison context remain attributes rather than automatically becoming new entity kinds.; source ids: part2-massspecgym-arxiv-v1; source locator: Table 4: Cosine similarity, formula; ambiguities: The formula/main label describes task input or candidate constraints in some MassSpecGym tables, not necessarily a different biological population. Preserve the paper-specific data identity and do not infer subset equivalence.