Table 3.
Loop number for CASP11, CASP12 and HOMSTRAD test sets.
| Size | CASP11 | CASP12 | HOMSTRAD | all |
|---|---|---|---|---|
| 69 (21) | 76 (18) | 23 | 168 (39) | |
| DaReUS-Loop | 67 (20) | 75 (17) | 23 | 165 (37) |
| NGK | 66 (18) | 76 (18) | 23 | 165 (36) |
| GalaxyLoop-PS2 | 50 (9) | 56 (9) | 19 | 125 (18) |
| Common ai | 47 | 55 | 19 | 121 |
| LoopIng | 63 (15) | 69 (13) | 22 | 154 (28) |
| Sphinx | 69 (21) | 76 (18) | 23 | 168 (39) |
| Common db | 62 | 69 | 22 | 153 |
| min(JSD) ≤ 0.2 | 53 (8) | 59 (6) | 23 | 135 (14) |
| NGK | 51 (6) | 59 (6) | 23 | 133 (12) |
| GalaxyLoop-PS2 | 40 (1) | 46 (4) | 19 | 105 (5) |
| CommonHC ai | 40 | 46 | 19 | 105 |
| LoopIng | 51 (6) | 55 (4) | 22 | 128 (10) |
| Sphinx | 53 (8) | 59 (6) | 23 | 135 (14) |
| CommonHC db | 51 | 55 | 22 | 128 |
size: number of loops identified. DaReUS-Loop, NGK, GalaxyLoop-PS2, LoopIng and Sphinx: number of loops that could be modeled using each approach. The number of long loops (at least 15 residues) are reported within parentheses. Common: number of target loops predictable by all different approaches, distinguishing ab initio (_ai) and data-based approaches (_db). CommonHC: subset of Common corresponding to loops predicted with a high confidence index (JSD leq 0.2, see below).