Table 2 |
||||
|
L/10 best scores accuracy improvements of GARP contact prediction for P2PconPred on the test set using parameters: 5% top scores a, W = 2, Geometric unweighted mutual coefficient. |
||||
|
Without GARP procedure |
With GARP procedure |
|||
|
|
||||
|
Accuracy protein |
Accuracy core |
Accuracy protein |
Accuracy core |
|
|
|
||||
|
P2PConPred |
0.12 ± 0.018 |
0.24 ± 0.027 |
0.18 ± 0.024 |
0.26 ± 0.030 |
|
PoCM |
0.23 ± 0.027 |
0.16 ± 0.021 |
- |
- |
|
|
||||
|
a In cases where less then L/10 predictions where reported by GARP, the calculation was repeated starting with the top 10% of scores. This was necessary for no more then five families in predicting contacts within the core. |
||||
|
Frenkel-Morgenstern et al. BMC Bioinformatics 2007 8(Suppl 5):S6 doi:10.1186/1471-2105-8-S5-S6 |
||||
