Proteins play an important role in structural constitution and functional activity of organisms.And all these take proteins forming certain spatial conformation as a premise.Amino acid sequences of proteins generated from evolution should contain the information for the assumption of the native secondary and tertiary structures.Furthermore, the functional role of proteins obliges it to keep thermostable and this thermostable information is encoded by protein sequences all the same.As a matter of fact, the fold-ability and the thermostability are both the characteristic properties of protein sequences.Elucidating these characteristic properties would contribute to protein structure predicting and new protein sequence designing.We study the folding related information in protein sequences by constructing aminoacid scoring matrix and modified statistical coupling analysis method (SCA).Herein, three protein structure class-specific score matrices (ALPHASUM, BETASUM and AFBETASUM) were constructed based on the structure alignment of low identity (<25%) all-alpha, all-beta, and alpha/beta proteins, respectively.The multiple sequence alignment of SH3 domain was analyzed using the method of SCA.The class-specific score matrices were significantly better than a structure-derived matrix (HSDM) and three other generalized matrices (BLOSUM30, BLOSUM60 and Gonnet250) in alignment performance tests.The optimized gap penalties also promote alignment performance.The statistical conserve energy from SCA method could evaluate the site conservation of SH3 sequence set properly.Several sites perturbing analysis revealed local and nonlocai perturbing modes in SH3 domain.Different perturbing modes which involved with different perturbing sites exist in SH3 domain.Through sharing common perturbing sites and responding sites, different perturbing modes could interact.And the coupling responding mode of all sites in the structure was determined.And sites with high average coupling energy correspond to those structural and functional important positions.In conclusion, different protein classes have distinct amino acid substitution patterns, and an amino acid score matrix should be constructed based on different structural classes.The coupling information about SH3 domain would improve our understanding about the relationship between protein sequence and its structure and function.It is also valuable in aiding new protein design .