The Research Of Mining Functional Modules In Uncertain Protein-Protein Interaction Network

Protein function module is one of the most basic units what comprise biological significance of cellular function and plays an important role in the process of combining various gene products. Then, using the theory of various disciplines of knowledge to detect protein function modules which are closely connected with biological function in protein interaction network becomes an important breakthrough for people to uncover the connection between protein interaction and biological function. Due to the particularity of biological process, the influence of factors such as the limitation of the experimental method, protein interaction data got by biological experiments is uncertainty. Therefore, an uncertain "possible semantic model" was proposed to deal with these uncertain data.This paper summarizes the existing research results. Based on the characteristics that protein interaction data obtained by biological experiment is often uncertaint and some existing weakness in current detecteding algorithms this paper put forward two new algorithms of detecting protein function modules based on uncertain network.An algorithm of OLHR is proposed to overcome the shortcomings that hierarchical clustering algorithm is weak in anti-noise and the classification results are non-overlapping. In OLHR, an indirect-connectivity value for proteion pairs is designed by analysising of the trend of human activities in social networking. Firstly, OLHR reconstruct the adjacency matrix of network, then using the existing modularity measure Q value and hierarchical clustering to partion network, finally, the preliminary results are extended to overlapping modules. The experimental results show that OLHR can find more meaningful protein modules.To improve the accuracy, this paper combine the GO terms and theory of uncertain graph and propose a "Core-Attachment algorithm based on expected-density and GO Terms to detect protein complexes in uncertain protein-protein interaction networks" named COMDG. The experimental results in two PPI networks of saccharomyces cerevisiae show that COMDG further improve the precision of detection for protein function modules.