Table 2. Detailed comparison between automatic and expert classifications.
Reference classification | Num. clust. | Homogeneity | Joining probability |
SCOP SF | 779 | 85.2 | 68.0 |
CATH SF | 885 | 81.1 | 66.4 |
SCOP or CATH SF | - | 86.3 | 69.1 |
SCOP folds | 466 | 92.0 | 44.5 |
CATH folds | 473 | 91.4 | 10.7 |
SCOP or CATH folds | - | 95.4 | 45.0 |
First column: reference classification. Second column: Number of clusters in the reference classification. Third column: Percentage of the 779 clusters in the automatic classification that are pure with respect to the reference classification (in case of CATH or SCOP, it is the fraction of clusters that are pure with respect to either CATH or SCOP). Fourth column: Percentage of the pairs joined in the reference classification that are joined in the automatic classification. In the case of folds, only pairs in different superfamilies are counted.