Sub-Network Identification In Complex Network Using Ranking And Rebuilding

DNA microarray technology is a high-throughput technology developing fast inrecent years. Microarray technology can measure thousands of genes at a time andgives their relative expression value. It has became an important method for bioinfoma-tric research. Comparing the microarray data of normal person and cancerous person,it is likely to discover genes those have diferent expression between these two condi-tions and help people understand the development of disease on gene level. How toextract genes that have diferential expression from mass of noise has been consideredby biologist, because most genes don’t express diferentially. It has been discoveredthat through adding prior information, noise can be reduced. Protein-protein interac-tion network describe the relationship between proteins. To find sub-network of PPIthat express diferentially under diferent conditions is a method to integrate prior in-formation. The popular method is define a score for every sub-network and search thesub-network which has the highest score, using simulated annealing or genetic algo-rithm. In this article, the author proposed a novel method to extract sub-network. The-oretical prove is given, along with plenty of simulation. The basic idea is to rank theedges of the network. The rank reflects the contribution the edge to the local network.Using this rank, the network is rebuilt, and the diferential expressing sub-network isextracted. Compared with popular methods, in stead of consider vertex or edge iso-lated, this method consider node vertex and edge in local network. The method isfaster. When extracting sub-network from PPI, the author taked the topology of thePPI network into account. Preliminary simulation results showed a better precisionand recall. More comprehensive conclusion needs further verification.