Table 1 |
||||
|
The prediction performances |
||||
|
Accuracy for interacting binding sites, also known as recall (%) |
Accuracy for non-interacting binding sites (%) |
Average accuracy (%) |
Precision (%) |
|
|
|
||||
|
Mix1 |
87.4 |
68.6 |
78.0 |
73.6 |
|
Mix1 (sequence window) |
88.7 |
63.3 |
76.0 |
70.8 |
|
Mix1 (putative binding sites) |
87.3 |
67.6 |
77.4 |
72.9 |
|
Homo-dimers2 |
96.4 |
67.5 |
81.9 |
74.8 |
|
Hetero-dimers2 |
66.3 |
62.8 |
64.5 |
64.0 |
|
|
||||
|
1 The mixed data set of homo-dimer and hetero-dimer interfaces. 2 The data set of homo-dimer and hetero-dimer interfaces were separately trained and tested. The prediction was performed using the default input features (sequence profile with a surface patch size 3). |
||||
|
Chung et al. BMC Bioinformatics 2007 8:223 doi:10.1186/1471-2105-8-223 |
||||
