We present and implement a distance-based clustering of amino acids within the framework of a statistically derived interaction matrix and show that the resulting groups faithfully reproduce, for well-designed sequences, thermodynamic stability in and kinetic accessibility to the native state. A simple interpretation of the groups is obtained by eigenanalysis of the interaction matrix.
@article{arxiv.cond-mat/0010244,
title = {Amino acid classes and the protein folding problem},
author = {Marek Cieplak and Neal S. Holter and Amos Maritan and Jayanth R. Banavar},
journal= {arXiv preprint arXiv:cond-mat/0010244},
year = {2009}
}
Comments
REVTeX, 11 pages, 4 figures, To appear in J. Chem. Phys