A method of Secondary Structure Prediction of Protein by Error Back Propagation

摘要

Protein structure analysis has been an important field in the bioinformatics in the post genome era. In the analysis of the protein, the secondary and tertiary structure prediction has been developed for almost a quarter of a century. But because of a lack of data the protein structure prediction only with the calculation of molecular level has a problem of time consuming and low accuracy. So it's necessary to use biological information in the prediction of protein structures. In the secondary structure analysis, predictions were performed on single sequences rather than families of homologous sequences, and there were relatively few known tertiary structures from which to derive parameters. In bioinformatics some researchers have been doing a good deal of works on protein structure analysis and prediction using neural network such as the Error Back Propagation. In this paper we implemented a method of protein secondary structure prediction by Error Back Propagation using only amino acid sequences and secondary structure information. We find out that the proposed system has an advantage in learning variable lengths of input patterns.