论文部分内容阅读
针对基因识别问题,基于DNA序列的3周期这一性质,首先给出了DNA序列功率和信噪比的快速算法并讨论了不同物种基因类型的阈值确定方法;在此基础上,建立了基于背景噪声抑制和频谱平滑的SNR频谱预处理模型,经过预处理后的频谱不仅大幅度抑制了背景噪声,同时保留了SNR频谱的模式特征.在编码序列识别上,对经典的EPND预测算法进行了改进,使用改进的EPND算法对经过预处理后频谱进行基因识别,实验结果显示这种基因识别模型具有优异的基因识别性能,比传统直接使用基于滑动窗口DFT的EPND识别算法在敏感度、特异性等评价指标上提高了2%-12%左右.
Aiming at the problem of gene recognition, based on the three cycles of DNA sequence, a fast algorithm of DNA sequence power and signal-to-noise ratio is presented and the threshold determination method of different genotypes is discussed. Based on this, Noise suppression and spectrum smoothing, the preprocessed spectrum not only greatly reduces the background noise but also preserves the pattern features of the SNR spectrum.On the basis of coding sequence recognition, the classical EPND prediction algorithm is improved , The improved EPND algorithm is used to identify the spectrum after the preprocessing. The experimental results show that this gene recognition model has excellent performance of gene recognition. Compared with the traditional EPND recognition algorithm based on sliding window DFT, the sensitivity, specificity, etc. Evaluation indicators increased by about 2% -12%.