Figure 4

Comparison of natural occurrences of 3-grams. Two subsets of 3-grams are compared (A) those located at (1) active sites of enzymes vs. (2) other positions and (B) those located at (1) active sites of enzymes vs. (2) exposed ones (see Methods). The 3-grams in an ensemble of 59 non-redundant enzyme/ligand complexes [24] were analyzed with regard to their natural frequencies of occurrence in the UniProt. The abscissa shows the number of occurrence of the examined 3-grams in UniProt, divided into intervals of size ΔCi = 1,000, and the ordinate shows the fraction of 3-grams fall in the successive count intervals, for the two subsets of 3-grams: active sites (gray) and others/exposed (black). Comparison of the two histograms reveals the higher proportion of rare 3-grams in the active sites. In particular the range C _i < 1,140 (rare 3-grams) shows a propensity for active site 3-grams that is enhanced by a factor of 2.67 compared to other 3-grams and by a factor of 5.4 compared to exposed 3-grams. Chi square test resulted with P << 0.001 for both histograms indicating the statistical significance of the differences.