基因识别问题及其算法实现  

Gene Identification Problem and Its Algorithm Implementation

作　　者：任红[1] 吴金龙[2] 罗莹[1] 

机构地区：[1]东南大学信息科学与工程学院,江苏南京210096 [2]东南大学能源与环境学院,江苏南京210096

出　　处：《数学的实践与认识》2013年第14期123-131,共9页Mathematics in Practice and Theory

摘　　要：主要针对基于频谱3-周期性的基因预测问题做了如下研究:提出了在Voss映射下功率谱与信噪比的快速算法及其与Z-curve映射的关系;给出了实数映射下功率谱与信噪比的快速计算公式;提出了分类阈值确定方法;设计了基于三层筛选的改进长度窗口滑动法和双向滑动的基于移动序列信噪比曲线的两种基因识别算法,提高了基因识别的准确率;最后提出了基因识别的另一特征指数——频谱相角并进行了相关分析.The paper mainly conducts the following research on the gene prediction problem based on the spectrum 3-base periodicity.A fast algorithm is proposed to calculate the power spectrum and the signal-to-noise ratio based on Voss mapping, and its relationship with Z- curve mapping is discussed. We also propose a fast algorithm about real mapping. Then, a classification threshold determination method is studied. Next, we propose two gene identi- fication algorithms. One is the improved length sliding window method based on the three layers screening and the other is the bidirectional slide method based on mobile sequence signal-to-noise ratio curve. Both the two methods are proved to be competent. In addi- tion, we propose and do some research on another characteristic index of gene identification problem spectrum phase angle.

关 键 词：三层筛选 双向滑动分类阈值 频谱相角 

分 类 号：TN911.4[电子电信—通信与信息系统] O242.1[电子电信—信息与通信工程]