Fig. 2

Cumulative binding residue similarities for FunFam and EC-number. The x-axis gives the fraction of binding residue annotations (Eq. 1) agreeing between all pairs of proteins in the same functional “groups” according to different sources: the fat black line marks the similarity within FunFams [3] and the gray fat line marks the similarity within same EC number [2]. For comparison the complements are also shown, namely the sub-sets of proteins in the same FunFam but with different EC number (dashed dark line) and in different FunFams but with the same EC (dashed gray line). All curves give reversely cumulative numbers answering the question: how many protein families had a binding residue annotation similarity (Eq. 1) above the similarity threshold shown on the x-axis? The two panels show the absolute count of protein families (a) and the fraction of all families (b) on the y-axis. For instance, 60% or more of all binding residues (indicated by rightmost vertical gray line; the middle vertical gray line marks the 50%) agreed within 354 FunFams (corresponding to 19%) and 145 identical EC numbers (corresponding to 14%). The leftmost vertical gray line marks random binding residue similarity (5.5 ± 0.2%). Contrary to all other groups, proteins grouped by the same EC number and differing FunFams (dashed gray line) have similarity scores close to random. The middle vertical gray lines mark the 50 and 60