| With the completion of the Human Genome Project, bioinformatics step into the post-genomic era. A key aim of post genomic biomedical research is to systematically catalogue all molecules and their interactions, which construct complex biological networks,within a living cell. Rapid advances in network biology indicate that cellular networks are governed by universal laws. Currently, one of the major challenge of bioinformatics is that discovery biological function from the topological structure of biological network.A new thinking of predicting network motif function is given in this thesis, based on the analysis of relativity among network motifs. Firstly, this thesis analysis the structure individual property of network motif based the sub graph analysis method. The property is benefit for the future research of network motif. Secondly, this thesis gives a new method to predict unknown-function network motif using the known-function network motif, based on the analysis of the interaction of network motifs. To two network motifs which contain same number of nodes, we analysis their function similarity based on the vector space model. To known function network motif with small size of nodes, a approach based on statistics is proposed to predict the function of network motif with big size of nodes and unknown function.This method is used to cluster the 5-size network motif in the E. coil gene regulate network and predict their function based on the feed forward loop network motif. In addition,this thesis first proposed a means to compare network motifs based on the vector space model and use k-means to cluster. Experimental results show that our model is available for the network motif function prediction. Finally, a network motif analysis of the system is given, which import the correlation degree algorithm to determine isomorphic. This greatly improves the efficiency of the system and benefit for the optimize of network motif detection algorithm.
|
PDF Full Text Request