Summary

We have developed an approach based on information theory to compute the structural information content, or structural entropy, of amino acid sequences of arbitrary lengths. The structural entropy of amino acid sequences is computed from their local structural distributions, using different structural representations. Structure entropy values could be used to indicate structure conservation quantitatively. As a practical tool in de novo protein design, our approach offers a quantitative measure of the effects of a single amino acid mutation on the change of local conformations. There are also qualitative correspondences between structure entropy values and proton exchange rates. Structure entropy gives more information than the primary sequence of protein, and provides insights to the study of protein local sequence/structure relationships