判别分析在基因剪切位点识别中的应用

Statistical Discriminant Analysis′ Application in the Identification of Gene′s Splice Sites

基因的结构预测是目前极为活跃的研究领域,而基因剪切位点的识别则是基因结构预测中的重要环节.本文采用自己提出的平均似然比等方法,同时采用目前较为有效的权重矩阵(WMM)、最大相关分解(MDD)、动态规划等模型,充分提取剪切位点信号区的统计特征,采用统计中判别分析的方法进行判别,得到了较好的判别效果,并与其它方法进行了比较.

The discrimination of splice sites is an important part in the process of gene＇s identification. During these years a lot of methods, such as probability model and mechine learning etc, are presented. In this paper a statistical method, discriminant analysis, is applied. We propose our new method based on average likelihood ratio when choosing the parameters. When dealing with the center of the signal region the Weight Matrix Model （WMM） and the Maximal Dependence Decomposition （MDD） procedure are also used. At last, after the calculation the results of discriminant for Acceptor site and Donor site are presented, compared with other methods a approving result is got.