三周期性是大多数基因组序列的编码区具有的主要特征.本文提出利用小波变换分析DNA序列编码区的三周期性,形成一种新的基于小波变换的DNA序列编码区预测方法,理论和实验研究证实了新方法的可行性,探测率和正确率分别达到81%和75%,特别是探测率较目前常用的其它一些方法有较大改善.
The major signal in protein coding regions for most of genomic sequences is three-base periodicity. In this paper, we analyze this periodicity using wavelet Wansformafion (WT) and propose a novel prediction approach for the protein coding regions of DNA sequences based on WT. This approach is able to predict and locate the coding regions simultaneously and is independent of training sets or existing database information. The validity of this approach is verified by a great deal of research results from theoretical analysis and experiments. The sensitivity and the specificity of novel approach reach 81% and 75 % respectively. So, the prediction effectiveness is good. Especially, the sensitivity of novel approach is greatly improved compared with other techniques currently in use.