Figure 4. Distributions of intra-superfamily and intra-cluster similarity scores.

(A) Distribution of the normalized total similarity score, Eq. (6) and (8), for domain pairs in the same superfamily. The grey bars are obtained for superfamilies that are not split, whereas the white bars are obtained for splitted superfamilies. One can see that splitted superfamilies present a bimodal distribution, with a peak with very small structure similarity. (B) Distribution of the mean intracluster similarity in the automatic classification, Eq. (11). The white bars are obtained for domains in clusters that contain only proteins of the same SCOP fold. The orange bars are obtained for minority domains in clusters containing domains that are mostly of a different SCOP fold.