基于傅里叶分析的蛋白质编码区预测中功率谱密度计算方法研究

Study on Power Spectrum Density Computation Methods for Protein Coding Regions Prediction

提出一种新的功率谱算法——多滑动窗周期图法.新算法和传统周期图功率谱密度算法在ALLSEQ和HMR195两个DNA序列集上进行了基因预测实验.以近似相关系数AC作为预测准确率测度,实验结果表明,新算法的预测准确率比基于传统周期图法的傅里叶分析基因预测算法有较大提高.

A widely used algorithm called traditional periodogram（TP） has been applied in experiments of prediction of protein coding regions.Then a new algorithm of power spectrum density（PSD） computation called multi-sliding-window periodogram（MSWP） is proposed and used in the experiments.Measured with approximate correlation（AC） for overall prediction accuracy,the new algorithm has outperformance comparing with the other TP based Fourier analysis gene prediction algorithms by taking HMR195 and ALLSEQ（also called BG570） as test datasets.