Discovering Motifs In Gene Sequences Using Networks

While gene is transcribing, it is always controlled by something called motifs which are some conservative fragments of DNA. And actually they make a great effect for the transcriptions of DNA. So how to discover these motifs from gene sequences has become one of the most important issues in current bioinformatics.There are many methods to discover motifs, and Gibbs Sampler Algorithm and EM Algorithm are considered to be the most effective and mature technology. However, there are some limitation of these traditional algorithms: for example, a lot of calculations may lead to just discovering finite motifs ;it lacks generality because some statistical models just suit for the motifs which are at certain conditions. In 2006, Jiang et al. present stochastic network model which combines EM algorithm and stochastic network and Frankin et al. present MotifCut which combined stochastic network with modified parametric flow algorithm in the same year.These algorithms overcome the limitations of traditional ones and get better outputs.The paper emphasizes the method of MotifCut algorithm , makes some improvements on the weight between nodes of the network ,uses the motifs in DNA which contains CRP binding site to compare with MotifCut .At last we prove our improvements can make the result much better.