Table 5.
The performance of various sequence-based domain prediction methods on the 97 (70 one-domain proteins and 27 multidomain proteins) CASP7 targets
| Methods | Domain number |
Domain position |
|||||
|---|---|---|---|---|---|---|---|
|
One |
Multi |
Combined |
Multi |
||||
| A | Sp | Sn | A | A | Complete | Partial | |
| FIEFDom | 100 | 88.9 | 30.8 | 29.6 | 64.8 | 6 | 2 |
| CHOP (25) | 55.8 | 37.5 | 42.9 | 25.0 | 40.4 | 4 | 4 |
| DomSSEA (28) | 92.9 | 100 | 30.8 | 30.8 | 61.8 | 4 | 4 |
| DPSa | 80.5 | 100 | 42.3 | 42.3 | 61.4 | 5 | 2 |
| HHPred1a | 95.6 | 100 | 25.9 | 25.9 | 60.8 | 4 | 3 |
| HHPred3a | 95.7 | 100 | 25.9 | 25.9 | 60.8 | 4 | 3 |
| NNPutLaba | 78.5 | 80.0 | 15.4 | 14.8 | 46.6 | 2 | 3 |
All values under the domain number prediction are percentages. Sequence-based domain prediction methods that were used in the CASP7 are listed on left. For one domain number prediction, the accuracy (A) is listed. For multidomain number prediction, accuracy (A), specificity (Sp) and sensitivity (Sn) are listed. The domain number prediction accuracy for all targets in CASP7 set is listed under the ‘Combined’ heading. For the domain position prediction of multidomain proteins, the actual count of the proteins whose domain boundaries are predicted completely correct and partially correct is listed.