中文摘要：

为对未知基因序列的编码区进行预测,利用基因编码区存在的频谱3-周期性质,在传统的固定窗口长度滑动窗算法的基础上,提出了一种改进型的基因识别算法.新算法将全相位谱分析技术与多采样率数字信号处理技术相结合,有效地降低了传统滑动窗算法中存在的截断效应,减少了计算量并且可实现流水线操作.最后通过计算机仿真将该算法所得结果与其他识别算法所得结果相比较,结果表明该算法在核苷酸水平上有较高的预测准确性.

英文摘要：

The 3-base periodicity, which exists in the coding regions, is adopted to predict the location of coding regions in an unknown DNA sequence. Based on the fixed-length sliding window approach, an improved algorithm for gene prediction is proposed, which combines all-phase FFT spectrum analysis and multi-rate digital signal processing. The method possesses some merits including reducing the truncation effect, improving the operation efficiency and providing a scheme for pipeline operation. Computer simulation results are given to compare the presented approach with other method for gene prediction, indicating that the proposed method achieves a satisfactory prediction accuracy at the nucleotide level.