An Approach To Gene Prediction Based On Mus Musculus Genes

With the completion of more and more species gene sequencing, particularly the sequencing of the human genome, the following problem is the interpretation and analysis of the genome. Faced with the huge genetic data, digging and analyzing life information of gene sequences is the most important issue in bioinformatics. Gene prediction is a significant task for DNA sequence analysis, while identifying coding regions is the key of gene prediction. Exon recognition of eukaryotic gene is of much difficulty in gene prediction. Due to the complex structure of eukaryotic gene coding region, existence of short exon sequences and a large number of repeated sequences and so on, the current prediction of the coding region of eukaryotes has not been able to achieve good results. In this article the study of eukaryotic gene prediction is proved by mouse genes.Three aspects of gene prediction are described in this article. Part Ⅰ:Introduction to bioinformatics, the situation and significance of gene prediction research at home and abroad. Part Ⅱ:The first step in gene prediction is the numerical mapping study of gene sequences. Introduce several statistical characteristics of gene sequences as well as extracts. Combine fisher Criterion on classification and statistical characteristics to classify gene sequences. Those researches pave the way for further research. Part Ⅲ:Research the power spectrum and SNR based on 3-periodicity, and select the appropriate threshold to discriminate. Experiment on mouse genes proved this conclusion that the power spectrum of mouse genes which is based on the argument has a higher peak in a third place in exon and is more stable in intron. Thereby, it increased the discrimination of the SNR between exons and introns. So the prediction of mouse gene performed a better result.In summary, the research of gene prediction algorithms in this article achieved better results on mouse genes whose average length is 150bp.