Table 3.
Comparison of the overall performance results using BaCelLo independent dataset (BaCelLo IDS)
| Predictor | Classes | Animals | Fungi | Plants |
| MultiLoc2-LowRes | 3 | 93 (93) | 79 (87) | 80 (83) |
| 4 | 80 (73) | 66 (60) | 72 (76) | |
| MultiLoc2-HighRes | 3 | 87 (89) | 71 (76) | 74 (71) |
| 4 | 75 (68) | 59 (52) | 65 (62) | |
| BaCelLo | 3 | 87 (91) | 88 (84) | 69 (76) |
| 4 | 69 (64) | 71 (57) | 61 (69) | |
| LOCtree | 3 | 76 (81) | 68 (75) | 73 (76) |
| 4 | 61 (62) | 55 (47) | 65 (70) | |
| Protein Prowler | 3 | 78 (91) | 75 (86) | 65 (63) |
| 4 | - | - | - | |
| TargetP | 3 | 86 (88) | 76 (82) | 72 (67) |
| 4 | - | - | - | |
| WoLF PSORT | 3 | 84 (88) | 77 (82) | 56 (69) |
| 4 | 69 (71) | 62 (51) | 46 (57) |
The average sensitivity and the overall accuracy (in parentheses) for the prediction of three and four classes for animals and fungi and four and five classes for plants are shown. Both measures are given in percentages. The top-scoring average sensitivity and average accuracy are highlighted in bold. Results for Protein Prowler and TargetP predictions are only available for a reduced number of classes since nu and cy are grouped as nu/cy.