Abstract:

Along with the progress of Human Genome Plan, people need to find out the genome structure from gained information. Recognition of splice sites is an essential part of this task. However, it is also a complicated problem, and satisfied results still cannot be reached. We design a splice site prediction algorithm based on BP network and SNP data to discover the influence of SNP on splice. In addition, a new encoding method is used to convert the DNA character sequences into decimal strings. Experimental results show the effectiveness of the SNP data and the new encoding method.

Key words: SNP;splice site prediction;artificial neural network;symbolic series analysis