An alphabet of amino acid conformations in protein

An approach to the representation of individual amino acids using dihedral angle measurements, based on the maxi-min distance algorithm presented by J.T. Tou and R.C. Gonzalez (1974), is outlined. The methodology, based on maxi-min and data set partitions, generates 124 distinct cluster centers from 19121 total residues. While only 18 of the 124 capture 1% or more of the data set, the first cis-proline conformation is 49th in population and suggests that over fifty centers may be significant. Four applications of these centers are demonstrated: representation of individual proteins, tripeptide sequence dependency, analysis of neighboring centers, and a simple structure prediction algorithm.<
>