Zhou G, Xu X, Zhang C T
Department of Physics, Tianjin University, People's Republic of China.
Eur J Biochem. 1992 Dec 15;210(3):747-9. doi: 10.1111/j.1432-1033.1992.tb17476.x.
A protein is generally classified into one of the following four structural classes: all alpha, all beta, alpha+beta and alpha/beta. In this paper, based on the weighting to the 20 constituent amino acids, a new method is proposed for predicting the structural class of a protein according to its amino acid composition. The 20 weighting parameters, which reflect the different properties of the 20 constituent amino acids, have been obtained from a training set of proteins through the linear-programming approach. The rate of correct prediction for a training set of proteins by means of the new method was 100%, whereas the highest rate of previous methods was 82.8%. Furthermore, the results showed that the more numerous training proteins, the more effective the new method.
