基因识别问题及其算法实现(上)

基因识别是目前生物信息学领域的一个研究热点,采用信号处理方法来识别基因编码序列也受到广泛重视.本文基于DNA序列的3-周期性,利用序列的频谱曲线和信噪比曲线,建立基因识别的简化算法.基于Voss映射,本文提出可变窗宽的DFT算法,代数推导出Voss映射和Z-curve映射下频谱和信噪比的函数关系.关于各类基因的阈值确定,本文运用统计学中的Bootstrap算法,取信噪比序列迭代所得均值的中位数,作为新的阈值,在精度指标下比较四组具有代表性的基因序列的阈值.进一步,运用本文提出算法对待测序列进行数据模型,确定编码区域.

Prediction of Gene Identification and its Algorithm Implementation

Gene identification is the current new hot point in the field of bioinformatics. It becomes more important to identify the protein coding regions with the signal processing method. This paper established the simplized algorithm of the gene identification by the curves of the frequency spectrum and noise-signal ratio sequences, which is based upon the 1/3 frequency feature. Basing on the Voss mapping, we provided the DFT algorithm with changing window width. Then we computed the functional relationship between the frequency spectrum and noise-signal ratio of the different Voss mapping and Z-curve mapping. As for the threshold value of the various gene sequences, we iterated the noise-signal ratio sequence and got the median of the mean value, then we chose the median value to be the updated threshold, which was the Bootstrap algorithm in statistics. With this method, we compared the four representative gene sequences with the updated threshold. Moreover, we computed to identify the provided gene sequences and provided their coding regions.