The Structural Classes of Proteins Predicted by Multi-resolution Analysis.

Prediction of protein structural class from primary structure is studied in this paper. Wavelet packet transform is used to decompose the corresponding numerical signal of protein into several sub-signals at different resolution scales. The auto-correlation functions based on the sub-signals are used as feature vectors of the protein. The Bayes decision rule is used as classification algorithm. Experiments show that for the same datasets, the prediction accuracy is improved compared with the existed methods. [ABSTRACT FROM AUTHOR]