Nakayama S, Shigezumi S, Yoshida M
University of Library and Information Science, Ibaraki, Japan.
J Chem Inf Comput Sci. 1988 May;28(2):72-8. doi: 10.1021/ci00058a006.
Proteins were represented as vectors, of which components were all possible pairs of amino acids. From a distance matrix between any pairs of proteins thus represented, several clusters corresponding to connected components were generated. Application of this method to three different sets of proteins showed that it was suitable for clustering closely related proteins with respect to the sequential similarity defined by Dayhoff.
