基于内插马尔可夫模型的Gibbs改进算法识别调控元件  

A Gibbs Sampling Algorithm Based on Interpolated Markov Model for Detecting Regulatory Elements

作　　者：谢雪英[1] 孙啸[1] 谢建明[1] 陆祖宏[1] 

机构地区：[1]东南大学吴健雄实验室,南京210096

出　　处：《中国生物医学工程学报》2006年第4期396-399,410,共5页Chinese Journal of Biomedical Engineering

基　　金：国家自然科学基金资助项目(60121101);"863"计划资助项目(2002AA231071);江苏省自然科学基金资助项目(BK2002057)

摘　　要：不同阶数插值形式的马尔可夫内插模型,可以表示在一个DNA序列中相邻核苷酸之间的前后关系的变化。本研究将内插马尔可夫模型引入Gibbs采样算法,识别基因上游序列中的调控元件。对模拟序列和10组来源于文献的酵母基因序列的测试结果表明,改进后的算法在识别保守性差的调控元件和抗噪声能力方面均优于传统的Gibbs采样算法。Interpolated Markov Model （IMM）, which combines several Markov models with different orders, can capture variable context dependencies between nearby nucleotides depending on the local composition of DNA sequence. Based on IMM, a polished Gibbs sampling algorithm has been developed in this work to detect the regulatory elements. Simulated data and real biological sequences from yeast Saccharomyces cerevisiae were used to test the polished algorithm. Results indicated that polished Gibbs sampling exhibited better performance in extracting the less-conserved elements and dealing with noisy sequences than single nucleotide independent model.

关 键 词：内插马尔可夫模型 GIBBS采样 调控元件 基因序列 

分 类 号：Q612[生物学—生物物理学]