Table 1.
Expert | Feature Source | Size | Coverage (%) |
P | HMS-PCI MS | 1 | 8.3 |
P | TAP MS | 1 | 8.8 |
P | Yeast-2-Hybrid | 1 | 3.9 |
F | GO Function | 21 | 80.7 |
F | GO Process | 33 | 76.1 |
F | GO Component | 23 | 81.5 |
F | Essentiality | 1 | 100 |
F | MIPS protein class | 25 | 4.6 |
F | MIPS mutant phenotype | 11 | 9.4 |
S | Gene fusion/cooccurence | 1 | 100 |
S | Sequence similarity | 1 | 100 |
S | Homology derived PPI | 4 | 100 |
S | Domain interaction | 1 | 100 |
E | Gene Expression | 20 | 88.9 |
E | Protein Expression | 1 | 42.8 |
E | Trans Factor Binding | 16 | 98.0 |
E | Synthetic Lethal | 1 | 7.6 |
Feature set derived for pairwise protein-protein interaction prediction in yeast. We used a total of 162 features from 17 different data sources. The first column lists the feature expert to which the feature source was assigned. We have designed a total of four experts: P, F, S and E (for definition see the 'Feature' section). The second column lists the name of the feature source. The third column lists the number of attributes from each source. The fourth column presents the average percentage of pairs for which information is available using this feature source. All related data sources and how they were converted into features have been described in details previous in: http://www.cs.cmu.edu/~qyj/papers_sulp/proteins05_pages/features.html.