基于不同算法的Motif预测比较分析与优化

Comparison, Analysis and Optimization of Motif Finding Based on Different Algorithms

研究转录因子结合位点(TFBs)的主要预测模型及其预测的算法,通过基于调控元件预测的3种代表性的算法MEME、Gibbs采样和Weeder预测拟南芥基因组。比较结果表明,Gibbs采样算法和Weeder算法预测长、短motif效率较高。重点分析MEME算法,提出结合不同算法查找motif的优化方法,并以实验验证该方法能有效提高预测效率。

This paper studies some models and discrimination algorithms of Transcription Factor Binding sites（TFBs）. Experiment compares advantages and disadvantages in three representative discrimination algorithms which are based on regulation elements, including MEME, Gibbs sample and Weeder through predicting arabidopsis thaliana genome, against Gibbs sampling algorithm and Weeder algorithms are forecast long and short motif of the characteristics of high efficiency, MEME is intensively analyzed, and proposed an effective way to forecast motifs through MEME binding other discrimination algorithms. Experimental result proves that the method can improve the efficiency of motif finding efficiently.