Amino acid compositions and evolutionary relationships with protein families.

We have examined the merits of the three functions based on amino acid compositions which have been proposed to indicate the similarity in amino acid sequences of two proteins; the difference index, the composition divergence and the composition coefficient. We have taken the amino acid compositions and sequences of 41 cytochrome c's and used the 820 values from all possible comparisons in the evaluation. We conclude that the functions do have a limited value in predicting proteins which are closely related in sequence and that the three functions are equivalent in this predictive ability. We have used the composition divergence values obtained from available pyruvate kinase amino acid compositions to generate a phylogenetic tree for this glycolytic enzyme.