Dependence of mutual information of big protein sequence

The mutual information function is used to describe the auto-correlation of amino acids in protein. We find two interesting phenomenon: (1) for any given big protein, the mutual information function l(k) is almost a const, where k is the length of gap. (2) for any two sequence similar proteins, the mutual information are nearly the same. As a consequent, we may use mutual information of protein as a character for sequences comparison. Foundation item: Supported by the National Natural Science Foundation of China (30170214) Biography: Shi Feng ( 1966-), male, Ph. D, Associate professor, research direction: bioinformatics.