Prediction Algorithms For Protein Interaction And Function Based On PPI Network

Both of the protein-protein interaction (PPI) prediction and the protein function prediction are the important research works in post-genomic era. Based on the PPI networks, there has been a remarkable line of research in the study of those two problems.Firstly, this paper summarized the research progress about the PPI prediction and protein function prediction. Then it analyzed the topological properties of the PPI networks and the distributed law of the false positives (FPs).According to the distribution of FPs in the PPI network which seems never considered by previous interaction prediction method, and the topological properties of PPI networks, we proposed an improved interaction prediction algorithm, VTC (Vertex to Clique). Compared with DC (defective clique) algorithm by experiments, VTC algorithm can predict protein interactions with not only higher reliability, but also larger quantity.In the PPI networks, the degree distribution follows a power law distribution. Moreover, the degree of most vertices is low. According to those characteristics, we have proposed a novel protein function prediction algorithm based on the Google search engine technology. In our method, we also utilize the preferential attachment criteria to improve the prediction reliability. Applied our method to S.cerevisiae PPI dataset, the experimental results have shown that our method has a high prediction performance.