用网络方法识别生物序列motif

生物序列motif的识别是后基因组时代的一个核心问题。本文首先回顾了识别motif的几种主要算法，然后根据motif的重要性和随机性介绍了利用网络识别motif的两种具有代表性的方法：一种是建立一个随机网络混合模型，利用EM算法识别其中随机的网络motif；另一种用修正的参数流算法过滤出其中的最大密度予图，即为生物序列motif，并指出这两种方法的优劣，最后还对今后研究方向给出了讨论。

Discovering Motifs using Networks

Motifs discovery is the core issue in the post- genomic period. The paper reviews several chief algorithms to discover motifs at first, then introduces two main representative algorithms of motifs discovery using networks to deal with the significations and randomness of networks and network motifs：one is to build a stochastic network to find out the stochastic network motifs with EM algorithm;the other is to filter the Maximum Density Subgraphs（MDS） in the network with the modified parametric flow algorithm. At last we point out the advantages and disadvantages both of them and propose the discussions for research for the future.