Table 1.
Retrieval statistics (per cent)
| Positive mode |
Negative mode |
|||||
|---|---|---|---|---|---|---|
| Baseline performance | +Element filter | +Known formula | Baseline performance | +Element filter | +Known formula | |
| Small database (HMDB, MassBank, ChEBI, NIST14) Correct in TOP 1 | ||||||
| FingerScorer | 37.5 ± 3.5 | 44.5 ± 3.5 (+7.0) | 46.0 ± 3.0 (+8.5) | 32.5 ± 4.5 | 37.5 ± 4.5 (+5.0) | 38.5 ± 4.5 (+6.0) |
| FragScorer | 50.0 ± 2.0 | 54.0 ± 2.0 (+4.0) | 56.0 ± 2.0 (+6.0) | 44.0 ± 3.0 | 45.5 ± 3.5 (+1.5) | 47.0 ± 4.0 (+3.0) |
| FingerScorer&FragScorer | 49.0 ± 2.0 | 52.0 ± 2.0 (+3.0) | 53.0 ± 2.0 (+4.0) | 42.5 ± 3.5 | 45.0 ± 3.0 (+2.5) | 46.0 ± 3.0 (+3.5) |
| MetFrag | 51.7 ± 0.0 | 51.7 ± 0.0 (+0.0) | 51.7 ± 0.0 (+0.0) | 45.0 ± 0.0 | 45.0 ± 0.0 (+0.0) | 45.0 ± 0.0 (+0.0) |
| CFM-IDa | 41.0 ± 4.0 | 47.5 ± 3.5 (+6.5) | 48.5 ± 2.5 (+7.5) | 36.5 ± 4.5 | 41.5 ± 3.5 (+5.0) | 42.5 ± 3.5 (+6.0) |
| CSI:FingerID | N/A | 42.5 ± 2.5 | 45.0 ± 3.0 (+2.5) | N/A | 38.0 ± 5.0 | 39.0 ± 5.0 (+1.0) |
| Small database (HMDB, MassBank, ChEBI, NIST) Correct in TOP 5 | ||||||
| FingerScorer | 80.0 ± 3.0 | 83.0 ± 3.0 (+3.0) | 83.5 ± 2.5 (+3.5) | 71.5 ± 5.5 | 74.5 ± 5.5 (+3.0) | 75.0 ± 5.0 (+3.5) |
| FragScorer | 82.5 ± 2.5 | 84.0 ± 2.0 (+1.5) | 84.5 ± 2.5 (+2.0) | 76.5 ± 4.5 | 77.5 ± 4.5 (+1.0) | 77.5 ± 4.5 (+1.0) |
| FingerScorer&FragScorer | 86.0 ± 2.0 | 87.0 ± 2.0 (+1.0) | 87.0 ± 2.0 (+1.0) | 78.0 ± 4.0 | 80.0 ± 4.0 (+2.0) | 80.0 ± 4.0 (+2.0) |
| MetFrag | 86.5 ± 0.0 | 88.6 ± 0.0 (+2.1) | 89.1 ± 0.0 (+2.6) | 83.5 ± 0.0 | 85.6 ± 0.0 (+2.1) | 85.6 ± 0.0 (+2.1) |
| CFM-IDa | 76.5 ± 3.5 | 80.5 ± 3.5 (+4.0) | 82.0 ± 2.0 (+5.5) | 68.0 ± 7.0 | 74.0 ± 2.0 (+6.0) | 74.0 ± 2.0 (+6.0) |
| CSI:FingerID | N/A | 76.0 ± 3.0 | 76.5 ± 2.5 (+0.5) | N/A | 69.0 ± 5.0 | 69.5 ± 4.5 (+0.5) |
| Large database (PubChem, HMDB, MassBank, ChEBI) Correct in TOP 20 | ||||||
| FingerScorer | 48.5 ± 4.5 | 52.0 ± 5.0 (+3.5) | 56.5 ± 5.5 (+8.0) | 47.0 ± 7.0 | 52.5 ± 8.5 (+5.5) | 56.0 ± 9.0 (+9.0) |
| FragScorer | 26.5 ± 1.5 | 34.5 ± 1.5 (+8.0) | 41.5 ± 1.5 (+15.0) | 36.5 ± 2.5 | 45.0 ± 4.0 (+8.5) | 45.5 ± 3.5 (+9.0) |
| FingerScorer&FragScorer | 45.0 ± 3.0 | 49.5 ± 3.5 (+4.5) | 52.5 ± 3.5 (+7.5) | 50.5 ± 7.5 | 56.0 ± 8.0 (+5.5) | 58.5 ± 8.5 (+8.0) |
| MetFrag | 32.8 ± 0.0 | 41.4 ± 0.0 (+8.6) | 50.0 ± 0.0 (+17.2) | 41.2 ± 0.0 | 48.5 ± 0.0 (+7.3) | 50.2 ± 0.0 (+9.0) |
| CFM-IDa | 23.0 ± 1.0 | 31.5 ± 1.5 (+8.5) | 37.5 ± 1.5 (+14.6) | 30.0 ± 0.0 | 34.5 ± 0.5 (+4.5) | 38.0 ± 1.0 (+8.0) |
| CSI:FingerID | N/A | 34.5 ± 4.5 | 38.0 ± 5.0 (+3.5) | N/A | 41.5 ± 6.5 | 43.5 ± 6.5 (+2.0) |
a
For CFM-ID the best retrieval rate is reported for all scoring functions used, however the best performing scoring function was not consistent between different tests (Supplementary Note S3). Improvement relative to baseline performance is shown in brackets.