Figure 2

(a) Distribution of similarity scores for different column types. (b) Distribution of correlation scores. (c) Distribution of ALLR scores. The distributions are based on the largest 100 families in SCOP 1.50 database. The pairs of profile columns are divided into five categories depending on the nature of the seed amino acids, as described in the text. Note that in general the distributions of correlation scores overlap more than the other scoring functions, and specifically, the overlap between the scores of identical columns and the scores of dissimilar columns is greater (24%) than the overlap between the same types of columns, using our similarity scores (2.1%) or the ALLR scores (4.4%).