A protein structural alphabet and its substitution matrix CLESUM
Abstract
Description
By using a mixture model for the density distribution of the three pseudobond angles formed by $C_α$ atoms of four consecutive residues, the local structural states are discretized as 17 conformational letters of a protein structural alphabet. This coarse-graining procedure converts a 3D structure to a 1D code sequence. A substitution matrix between these letters is constructed based on the structural alignments of the FSSP database.
10 pages
10 pages